[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-Guang000--Awesome-Dataset-Distillation":3,"tool-Guang000--Awesome-Dataset-Distillation":64},[4,17,27,35,43,56],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":16},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,3,"2026-04-05T11:01:52",[13,14,15],"开发框架","图像","Agent","ready",{"id":18,"name":19,"github_repo":20,"description_zh":21,"stars":22,"difficulty_score":23,"last_commit_at":24,"category_tags":25,"status":16},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",138956,2,"2026-04-05T11:33:21",[13,15,26],"语言模型",{"id":28,"name":29,"github_repo":30,"description_zh":31,"stars":32,"difficulty_score":23,"last_commit_at":33,"category_tags":34,"status":16},2271,"ComfyUI","Comfy-Org\u002FComfyUI","ComfyUI 是一款功能强大且高度模块化的视觉 AI 引擎，专为设计和执行复杂的 Stable Diffusion 图像生成流程而打造。它摒弃了传统的代码编写模式，采用直观的节点式流程图界面，让用户通过连接不同的功能模块即可构建个性化的生成管线。\n\n这一设计巧妙解决了高级 AI 绘图工作流配置复杂、灵活性不足的痛点。用户无需具备编程背景，也能自由组合模型、调整参数并实时预览效果，轻松实现从基础文生图到多步骤高清修复等各类复杂任务。ComfyUI 拥有极佳的兼容性，不仅支持 Windows、macOS 和 Linux 全平台，还广泛适配 NVIDIA、AMD、Intel 及苹果 Silicon 等多种硬件架构，并率先支持 SDXL、Flux、SD3 等前沿模型。\n\n无论是希望深入探索算法潜力的研究人员和开发者，还是追求极致创作自由度的设计师与资深 AI 绘画爱好者，ComfyUI 都能提供强大的支持。其独特的模块化架构允许社区不断扩展新功能，使其成为当前最灵活、生态最丰富的开源扩散模型工具之一，帮助用户将创意高效转化为现实。",107662,"2026-04-03T11:11:01",[13,14,15],{"id":36,"name":37,"github_repo":38,"description_zh":39,"stars":40,"difficulty_score":23,"last_commit_at":41,"category_tags":42,"status":16},3704,"NextChat","ChatGPTNextWeb\u002FNextChat","NextChat 是一款轻量且极速的 AI 助手，旨在为用户提供流畅、跨平台的大模型交互体验。它完美解决了用户在多设备间切换时难以保持对话连续性，以及面对众多 AI 模型不知如何统一管理的痛点。无论是日常办公、学习辅助还是创意激发，NextChat 都能让用户随时随地通过网页、iOS、Android、Windows、MacOS 或 Linux 端无缝接入智能服务。\n\n这款工具非常适合普通用户、学生、职场人士以及需要私有化部署的企业团队使用。对于开发者而言，它也提供了便捷的自托管方案，支持一键部署到 Vercel 或 Zeabur 等平台。\n\nNextChat 的核心亮点在于其广泛的模型兼容性，原生支持 Claude、DeepSeek、GPT-4 及 Gemini Pro 等主流大模型，让用户在一个界面即可自由切换不同 AI 能力。此外，它还率先支持 MCP（Model Context Protocol）协议，增强了上下文处理能力。针对企业用户，NextChat 提供专业版解决方案，具备品牌定制、细粒度权限控制、内部知识库整合及安全审计等功能，满足公司对数据隐私和个性化管理的高标准要求。",87618,"2026-04-05T07:20:52",[13,26],{"id":44,"name":45,"github_repo":46,"description_zh":47,"stars":48,"difficulty_score":23,"last_commit_at":49,"category_tags":50,"status":16},2268,"ML-For-Beginners","microsoft\u002FML-For-Beginners","ML-For-Beginners 是由微软推出的一套系统化机器学习入门课程，旨在帮助零基础用户轻松掌握经典机器学习知识。这套课程将学习路径规划为 12 周，包含 26 节精炼课程和 52 道配套测验，内容涵盖从基础概念到实际应用的完整流程，有效解决了初学者面对庞大知识体系时无从下手、缺乏结构化指导的痛点。\n\n无论是希望转型的开发者、需要补充算法背景的研究人员，还是对人工智能充满好奇的普通爱好者，都能从中受益。课程不仅提供了清晰的理论讲解，还强调动手实践，让用户在循序渐进中建立扎实的技能基础。其独特的亮点在于强大的多语言支持，通过自动化机制提供了包括简体中文在内的 50 多种语言版本，极大地降低了全球不同背景用户的学习门槛。此外，项目采用开源协作模式，社区活跃且内容持续更新，确保学习者能获取前沿且准确的技术资讯。如果你正寻找一条清晰、友好且专业的机器学习入门之路，ML-For-Beginners 将是理想的起点。",84991,"2026-04-05T10:45:23",[14,51,52,53,15,54,26,13,55],"数据工具","视频","插件","其他","音频",{"id":57,"name":58,"github_repo":59,"description_zh":60,"stars":61,"difficulty_score":10,"last_commit_at":62,"category_tags":63,"status":16},3128,"ragflow","infiniflow\u002Fragflow","RAGFlow 是一款领先的开源检索增强生成（RAG）引擎，旨在为大语言模型构建更精准、可靠的上下文层。它巧妙地将前沿的 RAG 技术与智能体（Agent）能力相结合，不仅支持从各类文档中高效提取知识，还能让模型基于这些知识进行逻辑推理和任务执行。\n\n在大模型应用中，幻觉问题和知识滞后是常见痛点。RAGFlow 通过深度解析复杂文档结构（如表格、图表及混合排版），显著提升了信息检索的准确度，从而有效减少模型“胡编乱造”的现象，确保回答既有据可依又具备时效性。其内置的智能体机制更进一步，使系统不仅能回答问题，还能自主规划步骤解决复杂问题。\n\n这款工具特别适合开发者、企业技术团队以及 AI 研究人员使用。无论是希望快速搭建私有知识库问答系统，还是致力于探索大模型在垂直领域落地的创新者，都能从中受益。RAGFlow 提供了可视化的工作流编排界面和灵活的 API 接口，既降低了非算法背景用户的上手门槛，也满足了专业开发者对系统深度定制的需求。作为基于 Apache 2.0 协议开源的项目，它正成为连接通用大模型与行业专有知识之间的重要桥梁。",77062,"2026-04-04T04:44:48",[15,14,13,26,54],{"id":65,"github_repo":66,"name":67,"description_en":68,"description_zh":69,"ai_summary_zh":69,"readme_en":70,"readme_zh":71,"quickstart_zh":72,"use_case_zh":73,"hero_image_url":74,"owner_login":75,"owner_name":76,"owner_avatar_url":77,"owner_bio":78,"owner_company":79,"owner_location":80,"owner_email":81,"owner_twitter":81,"owner_website":82,"owner_url":83,"languages":84,"stars":101,"forks":102,"last_commit_at":103,"license":104,"difficulty_score":105,"env_os":106,"env_gpu":107,"env_ram":107,"env_deps":108,"category_tags":111,"github_topics":112,"view_count":23,"oss_zip_url":81,"oss_zip_packed_at":81,"status":16,"created_at":115,"updated_at":116,"faqs":117,"releases":148},1791,"Guang000\u002FAwesome-Dataset-Distillation","Awesome-Dataset-Distillation","A curated list of awesome papers on dataset distillation and related applications.","Awesome-Dataset-Distillation 是一个专注于“数据集蒸馏”领域的精选资源库，汇集了该方向最全面、前沿的学术论文与应用案例。简单来说，它的核心目标是帮助研究者从海量原始数据中提炼出一个极小的合成数据集，使得模型仅在这个小数据集上训练，就能达到在原始大数据集上训练的同等高性能。\n\n这一工具主要解决了大模型训练中数据存储成本高、计算资源消耗大以及隐私保护难等痛点。通过数据集蒸馏技术，用户不仅能大幅降低训练门槛，还能在持续学习、神经架构搜索及数据隐私保护等场景中实现更高效的应用。\n\nAwesome-Dataset-Distillation 特别适合人工智能研究人员、算法工程师及相关领域的开发者使用。它由该领域的三位奠基人共同维护，不仅系统梳理了从 2018 年概念提出至今的技术演进脉络，还实时收录了包括梯度匹配、时间序列压缩及推荐系统应用在内的最新突破。无论是希望快速入门的新手，还是寻求前沿灵感的资深专家，都能在这里找到极具价值的参考文献与代码资源，是探索数据高效利用不可或缺的指南。","# Awesome Dataset Distillation \n\n[![Awesome](https:\u002F\u002Fcdn.rawgit.com\u002Fsindresorhus\u002Fawesome\u002Fd7305f38d29fed78fa85652e3a63e154dd8e8829\u002Fmedia\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fsindresorhus\u002Fawesome)\n\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FContributions-Welcome-278ea5\" alt=\"Contrib\"\u002F> \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FNumber%20of%20Items-321-FF6F00\" alt=\"PaperNum\"\u002F> ![Stars](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002FGuang000\u002FAwesome-Dataset-Distillation?color=yellow&label=Stars) ![Forks](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fforks\u002FGuang000\u002FAwesome-Dataset-Distillation?color=green&label=Forks)\n\n**Awesome Dataset Distillation** provides the most comprehensive and detailed information on the Dataset Distillation field.\n\n**Dataset distillation** is the task of synthesizing a small dataset such that models trained on it achieve high performance on the original large dataset. A dataset distillation algorithm takes as **input** a large real dataset to be distilled (training set), and **outputs** a small synthetic distilled dataset, which is evaluated via testing models trained on this distilled dataset on a separate real dataset (validation\u002Ftest set). A good small distilled dataset is not only useful in dataset understanding, but has various applications (e.g., continual learning, privacy, neural architecture search, etc.). This task was first introduced in the paper [*Dataset Distillation* [Tongzhou Wang et al., '18]](https:\u002F\u002Fwww.tongzhouwang.info\u002Fdataset_distillation\u002F), along with a proposed algorithm using backpropagation through optimization steps. Then the task was first extended to the real-world datasets in the paper [*Medical Dataset Distillation* [Guang Li et al., '19]](https:\u002F\u002Farxiv.org\u002Fabs\u002F2104.02857), which also explored the privacy preservation possibilities of dataset distillation. In the paper [*Dataset Condensation* [Bo Zhao et al., '20]](https:\u002F\u002Farxiv.org\u002Fabs\u002F2006.05929), gradient matching was first introduced and greatly promoted the development of the dataset distillation field.\n\nIn recent years (2022-now), dataset distillation has gained increasing attention in the research community, across many institutes and labs. More papers are now being published each year. These wonderful researches have been constantly improving dataset distillation and exploring its various variants and applications.\n\n**This project is curated and maintained by [Guang Li](https:\u002F\u002Fwww-lmd.ist.hokudai.ac.jp\u002Fmember\u002Fguang-li\u002F), [Bo Zhao](https:\u002F\u002Fwww.bozhao.me\u002F), and [Tongzhou Wang](https:\u002F\u002Fwww.tongzhouwang.info\u002F).**\n\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FGuang000_Awesome-Dataset-Distillation_readme_e5c26c8ba5f0.jpg\" width=\"20%\"\u002F>\n\n#### [How to submit a pull request?](.\u002FCONTRIBUTING.md)\n\n+ :globe_with_meridians: Project Page\n+ :octocat: Code\n+ :book: `bibtex`\n\n## Latest Updates\n+ [2026\u002F04\u002F02] [Learnability-Guided Diffusion for Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2604.00519) (Jeffrey A. Chan-Santiago et al., CVPR 2026) [:globe_with_meridians:](https:\u002F\u002Fjachansantiago.com\u002Flearnability-guided-distillation\u002F) [:book:](.\u002Fcitations\u002Fchansantiago2026learnability.txt)\n+ [2026\u002F03\u002F26] [FD2: A Dedicated Framework for Fine-Grained Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.25144) (Hongxu Ma & Guang Li et al., 2026) [:book:](.\u002Fcitations\u002Fma2026fd2.txt)\n+ [2026\u002F03\u002F26] [DIET: Learning to Distill Dataset Continually for Recommender Systems](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.24958) (Jiaqing Zhang et al., 2026) [:book:](.\u002Fcitations\u002Fzhang2026diet.txt)\n+ [2026\u002F03\u002F26] [Dataset Distillation Efficiently Encodes Low-Dimensional Representations from Gradient-Based Learning of Non-Linear Tasks](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.14830) (Yuri Kinoshita et al., 2026) [:book:](.\u002Fcitations\u002Fkinoshita2026lowdim.txt)\n+ [2026\u002F03\u002F26] [ShapeCond: Fast Shapelet-Guided Dataset Condensation for Time Series Classification](https:\u002F\u002Farxiv.org\u002Fabs\u002F2602.09008) (Sijia Peng et al., 2026) [:octocat:](https:\u002F\u002Fgithub.com\u002Flunaaa95\u002FShapeCond) [:book:](.\u002Fcitations\u002Fpeng2026shapecond.txt)\n+ [2026\u002F03\u002F26] [PRISM: Diversifying Dataset Distillation by Decoupling Architectural Priors](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.09905) (Brian B. Moser et al., TMLR 2026) [:octocat:](https:\u002F\u002Fgithub.com\u002FBrian-Moser\u002Fprism) [:book:](.\u002Fcitations\u002Fmoser2026prism.txt)\n+ [2026\u002F03\u002F23] [IMS3: Breaking Distributional Aggregation in Diffusion-Based Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.13960) (Chenru Wang & Yunyi Chen et al., CVPR 2026) [:book:](.\u002Fcitations\u002Fwang2026ims3.txt)\n+ [2026\u002F03\u002F23] [EVLF: Early Vision-Language Fusion for Generative Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.07476) (Wenqi Cai et al., CVPR 2026) [:globe_with_meridians:](https:\u002F\u002Fwenqi-cai297.github.io\u002Fearlyfusion-HP\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fwenqi-cai297\u002Fearlyfusion-for-dd\u002F) [:book:](.\u002Fcitations\u002Fcai2026evlf.txt)\n+ [2026\u002F03\u002F23] [HIERAMP: Coarse-to-Fine Autoregressive Amplification for Generative Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.06932) (Lin Zhao & Xinru Jiang et al., CVPR 2026) [:octocat:](https:\u002F\u002Fgithub.com\u002FOshikaka\u002FHIERAMP) [:book:](.\u002Fcitations\u002Fzhao2026hieramp.txt)\n+ [2026\u002F03\u002F06] [UniRain: Unified Image Deraining with RAG-based Dataset Distillation and Multi-objective Reweighted Optimization](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.03967) (Qianfeng Yang et al., CVPR 2026) [:octocat:](https:\u002F\u002Fgithub.com\u002FQianfengY\u002FUniRain) [:book:](.\u002Fcitations\u002Fyang2026unirain.txt)\n+ [2026\u002F03\u002F06] [Fixed Anchors Are Not Enough: Dynamic Retrieval and Persistent Homology for Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2602.24144) (Muquan Li et al., CVPR 2026) [:book:](.\u002Fcitations\u002Fli2026reta.txt)\n+ [2026\u002F03\u002F06] [ManifoldGD: Training-Free Hierarchical Manifold Guidance for Diffusion-Based Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2602.23295) (Ayush Roy et al., CVPR 2026) [:octocat:](https:\u002F\u002Fgithub.com\u002FAyushRoy2001\u002FManifoldGD) [:book:](.\u002Fcitations\u002Froy2026manifold.txt)\n+ [2026\u002F03\u002F06] [PRISM: Video Dataset Condensation with Progressive Refinement and Insertion for Sparse Motion](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.22564) (Jaehyun Choi et al., CVPR 2026) [:book:](.\u002Fcitations\u002Fchoi2026prism.txt)\n\n## Contents\n- [Main](#main)\n  - [Early Work](#early-work)\n  - [Gradient\u002FTrajectory Matching Surrogate Objective](#gradient-objective)\n  - [Distribution\u002FFeature Matching Surrogate Objective](#feature-objective)\n  - [Kernel-Based Distillation](#kernel)\n  - [Distilled Dataset Parametrization](#parametrization)\n  - [Generative Distillation](#generative)\n  - [Better Optimization](#optimization)\n  - [Better Understanding](#understanding)\n  - [Label Distillation](#label)\n  - [Dataset Quantization](#quant)\n  - [Decoupled Distillation](#decouple)\n  - [Multimodal Distillation](#multi)\n  - [Self-Supervised Distillation](#self)\n  - [Benchmark](#benchmark)\n  - [Survey](#survey)\n  - [Ph.D. Thesis](#thesis)\n  - [Workshop](#workshop)\n  - [Challenge](#challenge)\n- [Applications](#applications)\n  - [Continual Learning](#continual)\n  - [Privacy](#privacy)\n  - [Medical](#medical)\n  - [Federated Learning](#fed)\n  - [Graph Neural Network](#gnn)\n  - [Neural Architecture Search](#nas)\n  - [Fashion, Art, and Design](#fashion)\n  - [Recommender Systems](#rec)\n  - [Blackbox Optimization](#blackbox)\n  - [Robustness](#robustness)\n  - [Fairness](#fairness)\n  - [Text](#text)\n  - [Video](#video)\n  - [Tabular](#tabular)\n  - [Retrieval](#retrieval)\n  - [Domain Adaptation](#domain)\n  - [Super Resolution](#super)\n  - [Time Series](#time)\n  - [Speech](#speech)\n  - [Machine Unlearning](#unlearning)\n  - [Reinforcement Learning](#rl)\n  - [Long-Tail](#long)\n  - [Learning with Noisy Labels](#noisy)\n  - [Object Detection](#detection)\n  - [Point Cloud](#point)\n  - [Universal Distillation](#uni)\n  - [Spiking Neural Network](#snn)\n  - [EEG](#eeg)\n  - [Finance](#finance)\n  - [Music](#music)\n  - [Remote Sensing](#rs)\n  - [Deraining](#dr)\n  - [Fine-Grained](#fine)\n\u003Ca name=\"main\" \u002F>\n\n## Main\n+ [Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F1811.10959) (Tongzhou Wang et al., 2018) [:globe_with_meridians:](https:\u002F\u002Fssnl.github.io\u002Fdataset_distillation\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002FSsnL\u002Fdataset-distillation) [:book:](.\u002Fcitations\u002Fwang2018datasetdistillation.txt)\n\n\u003Ca name=\"early-work\" \u002F>\n\n### Early Work\n+ [Gradient-Based Hyperparameter Optimization Through Reversible Learning](https:\u002F\u002Farxiv.org\u002Fabs\u002F1502.03492) (Dougal Maclaurin et al., ICML 2015) [:octocat:](https:\u002F\u002Fgithub.com\u002FHIPS\u002Fhypergrad) [:book:](.\u002Fcitations\u002Fmaclaurin2015gradient.txt)\n\n\u003Ca name=\"gradient-objective\" \u002F>\n\n### Gradient\u002FTrajectory Matching Surrogate Objective\n+ [Dataset Condensation with Gradient Matching](https:\u002F\u002Farxiv.org\u002Fabs\u002F2006.05929) (Bo Zhao et al., ICLR 2021) [:octocat:](https:\u002F\u002Fgithub.com\u002FVICO-UoE\u002FDatasetCondensation) [:book:](.\u002Fcitations\u002Fzhao2021datasetcondensation.txt)\n+ [Dataset Condensation with Differentiable Siamese Augmentation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2102.08259) (Bo Zhao et al., ICML 2021) [:octocat:](https:\u002F\u002Fgithub.com\u002FVICO-UoE\u002FDatasetCondensation) [:book:](.\u002Fcitations\u002Fzhao2021differentiatble.txt)\n+ [Dataset Distillation by Matching Training Trajectories](https:\u002F\u002Farxiv.org\u002Fabs\u002F2203.11932) (George Cazenavette et al., CVPR 2022) [:globe_with_meridians:](https:\u002F\u002Fgeorgecazenavette.github.io\u002Fmtt-distillation\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fgeorgecazenavette\u002Fmtt-distillation) [:book:](.\u002Fcitations\u002Fcazenavette2022dataset.txt)\n+ [Dataset Condensation with Contrastive Signals](https:\u002F\u002Farxiv.org\u002Fabs\u002F2202.02916) (Saehyung Lee et al., ICML 2022) [:octocat:](https:\u002F\u002Fgithub.com\u002Fsaehyung-lee\u002Fdcc) [:book:](.\u002Fcitations\u002Flee2022dataset.txt)\n+ [Loss-Curvature Matching for Dataset Selection and Condensation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2303.04449) (Seungjae Shin & Heesun Bae et al., AISTATS 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002FSJShin-AI\u002FLCMat) [:book:](.\u002Fcitations\u002Fshin2023lcmat.txt)\n+ [Minimizing the Accumulated Trajectory Error to Improve Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2211.11004) (Jiawei Du & Yidi Jiang et al., CVPR 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002FAngusDujw\u002FFTD-distillation) [:book:](.\u002Fcitations\u002Fdu2023minimizing.txt)\n+ [Scaling Up Dataset Distillation to ImageNet-1K with Constant Memory](https:\u002F\u002Farxiv.org\u002Fabs\u002F2211.10586) (Justin Cui et al., ICML 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Fjustincui03\u002Ftesla) [:book:](.\u002Fcitations\u002Fcui2022scaling.txt) \n+ [Sequential Subset Matching for Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2311.01570) (Jiawei Du et al., NeurIPS 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Fshqii1j\u002Fseqmatch) [:book:](.\u002Fcitations\u002Fdu2023seqmatch.txt)\n+ [Towards Lossless Dataset Distillation via Difficulty-Aligned Trajectory Matching](https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.05773) (Ziyao Guo & Kai Wang et al., ICLR 2024) [:globe_with_meridians:](https:\u002F\u002Fgzyaftermath.github.io\u002FDATM\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002FGzyAftermath\u002FDATM) [:book:](.\u002Fcitations\u002Fguo2024datm.txt)\n+ [SelMatch: Effectively Scaling Up Dataset Distillation via Selection-Based Initialization and Partial Updates by Trajectory Matching](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.18561) (Yongmin Lee et al., ICML 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FYongalls\u002FSelMatch) [:book:](.\u002Fcitations\u002Flee2024selmatch.txt)\n+ [Dataset Distillation by Automatic Training Trajectories](https:\u002F\u002Farxiv.org\u002Fabs\u002F2407.14245) (Dai Liu et al., ECCV 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FNiaLiu\u002FATT) [:book:](.\u002Fcitations\u002Fliu2024att.txt)\n+ [Neural Spectral Decomposition for Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2408.16236) (Shaolei Yang et al., ECCV 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fslyang2021\u002FNSD) [:book:](.\u002Fcitations\u002Fyang2024nsd.txt)\n+ [Prioritize Alignment in Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2408.03360) (Zekai Li & Ziyao Guo et al., 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FNUS-HPC-AI-Lab\u002FPAD) [:book:](.\u002Fcitations\u002Fli2024pad.txt)\n+ [Towards Stable and Storage-efficient Dataset Distillation: Matching Convexified Trajectory](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.19827) (Wenliang Zhong et al., CVPR 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002FZhong0x29a\u002FMCT) [:book:](.\u002Fcitations\u002Fzhong2025mct.txt)\n+ [Emphasizing Discriminative Features for Dataset Distillation in Complex Scenarios](https:\u002F\u002Farxiv.org\u002Fabs\u002F2410.17193) (Kai Wang & Zekai Li et al., CVPR 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002FNUS-HPC-AI-Lab\u002FEDF) [:book:](.\u002Fcitations\u002Fwang2025edf.txt)\n\n\u003Ca name=\"feature-objective\" \u002F>\n\n### Distribution\u002FFeature Matching Surrogate Objective\n+ [CAFE: Learning to Condense Dataset by Aligning Features](https:\u002F\u002Farxiv.org\u002Fabs\u002F2203.01531) (Kai Wang & Bo Zhao et al., CVPR 2022) [:octocat:](https:\u002F\u002Fgithub.com\u002Fkaiwang960112\u002Fcafe) [:book:](.\u002Fcitations\u002Fwang2022cafe.txt)\n+ [Dataset Condensation with Distribution Matching](https:\u002F\u002Farxiv.org\u002Fabs\u002F2110.04181) (Bo Zhao et al., WACV 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002FVICO-UoE\u002FDatasetCondensation) [:book:](.\u002Fcitations\u002Fzhao2023distribution.txt)\n+ [Improved Distribution Matching for Dataset Condensation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2307.09742) (Ganlong Zhao et al., CVPR 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Fuitrbn\u002FIDM) [:book:](.\u002Fcitations\u002Fzhao2023idm.txt)\n+ [DataDAM: Efficient Dataset Distillation with Attention Matching](https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.00093) (Ahmad Sajedi & Samir Khaki et al., ICCV 2023) [:globe_with_meridians:](https:\u002F\u002Fdatadistillation.github.io\u002FDataDAM\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002FDataDistillation\u002FDataDAM) [:book:](.\u002Fcitations\u002Fsajedi2023datadam.txt)\n+ [M3D: Dataset Condensation by Minimizing Maximum Mean Discrepancy](https:\u002F\u002Farxiv.org\u002Fabs\u002F2312.15927) (Hansong Zhang & Shikun Li et al., AAAI 2024)  [:octocat:](https:\u002F\u002Fgithub.com\u002FHansong-Zhang\u002FM3D) [:book:](.\u002Fcitations\u002Fzhang2024m3d.txt)\n+ [Exploiting Inter-sample and Inter-feature Relations in Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2404.00563) (Wenxiao Deng et al., CVPR 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FVincenDen\u002FIID) [:book:](.\u002Fcitations\u002Fdeng2024iid.txt)\n+ [Dataset Condensation with Latent Quantile Matching](https:\u002F\u002Fopenaccess.thecvf.com\u002Fcontent\u002FCVPR2024W\u002FDDCV\u002Fhtml\u002FWei_Dataset_Condensation_with_Latent_Quantile_Matching_CVPRW_2024_paper.html) (Wei Wei et al., CVPR 2024 Workshop) [:book:](.\u002Fcitations\u002Fwei2024lqm.txt)\n+ [DANCE: Dual-View Distribution Alignment for Dataset Condensation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.01063) (Hansong Zhang et al., IJCAI 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FHansong-Zhang\u002FDANCE) [:book:](.\u002Fcitations\u002Fzhang2024dance.txt)\n+ [Diversified Semantic Distribution Matching for Dataset Distillation](https:\u002F\u002Fdl.acm.org\u002Fdoi\u002F10.1145\u002F3664647.3680900) (Hongcheng Li et al., MM 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FLi-Hongcheng\u002FDSDM) [:book:](.\u002Fcitations\u002Fli2024dsdm.txt)\n+ [Dataset Distillation with Neural Characteristic Function: A Minmax Perspective](https:\u002F\u002Farxiv.org\u002Fabs\u002F2502.20653) (Shaobo Wang et al., CVPR 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002Fgszfwsb\u002FNCFM) [:book:](.\u002Fcitations\u002Fwang2025ncfm.txt)\n+ [OPTICAL: Leveraging Optimal Transport for Contribution Allocation in Dataset Distillation](https:\u002F\u002Fopenaccess.thecvf.com\u002Fcontent\u002FCVPR2025\u002Fhtml\u002FCui_OPTICAL_Leveraging_Optimal_Transport_for_Contribution_Allocation_in_Dataset_Distillation_CVPR_2025_paper.html) (Xiao Cui et al., CVPR 2025) [:book:](.\u002Fcitations\u002Fcui2025optical.txt)\n+ [Dataset Distillation via the Wasserstein Metric](https:\u002F\u002Farxiv.org\u002Fabs\u002F2311.18531) (Haoyang Liu et al., ICCV 2025) [:globe_with_meridians:](https:\u002F\u002Fliu-hy.github.io\u002FWMDD\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002FLiu-Hy\u002FWMDD) [:book:](.\u002Fcitations\u002Fliu2025wasserstein.txt)\n+ [Diversity-Enhanced Distribution Alignment for Dataset Distillation](https:\u002F\u002Fopenaccess.thecvf.com\u002Fcontent\u002FICCV2025\u002Fhtml\u002FLi_Diversity-Enhanced_Distribution_Alignment_for_Dataset_Distillation_ICCV_2025_paper.html) (Hongcheng Li et al., ICCV 2025) [:book:](.\u002Fcitations\u002Fli2025deda.txt)\n+ [Hyperbolic Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.24623) (Wenyuan Li & Guang Li et al., NeurIPS 2025) [:globe_with_meridians:](https:\u002F\u002Fguang000.github.io\u002FHDD-Webpage\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002FGuang000\u002FHDD) [:book:](.\u002Fcitations\u002Fli2025hdd.txt)\n+ [TGDD: Trajectory Guided Dataset Distillation with Balanced Distribution](https:\u002F\u002Farxiv.org\u002Fabs\u002F2512.02469) (Fengli Ran et al., AAAI 2026) [:octocat:](https:\u002F\u002Fgithub.com\u002FFlyFinley\u002FTGDD) [:book:](.\u002Fcitations\u002Fran2026tgdd.txt)\n \n\u003Ca name=\"kernel\" \u002F>\n\n### Kernel-Based Distillation\n+ [Dataset Meta-Learning from Kernel Ridge-Regression](https:\u002F\u002Farxiv.org\u002Fabs\u002F2011.00050) (Timothy Nguyen et al., ICLR 2021) [:octocat:](https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fneural-tangents) [:book:](.\u002Fcitations\u002Fnguyen2021kip.txt)\n+ [Dataset Distillation with Infinitely Wide Convolutional Networks](https:\u002F\u002Farxiv.org\u002Fabs\u002F2107.13034) (Timothy Nguyen et al., NeurIPS 2021) [:octocat:](https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fneural-tangents) [:book:](.\u002Fcitations\u002Fnguyen2021kipimprovedresults.txt)\n+ [Dataset Distillation using Neural Feature Regression](https:\u002F\u002Farxiv.org\u002Fabs\u002F2206.00719) (Yongchao Zhou et al., NeurIPS 2022) [:globe_with_meridians:](https:\u002F\u002Fsites.google.com\u002Fview\u002Ffrepo) [:octocat:](https:\u002F\u002Fgithub.com\u002Fyongchao97\u002FFRePo) [:book:](.\u002Fcitations\u002Fzhou2022dataset.txt)\n+ [Efficient Dataset Distillation using Random Feature Approximation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2210.12067) (Noel Loo et al., NeurIPS 2022) [:octocat:](https:\u002F\u002Fgithub.com\u002Fyolky\u002FRFAD) [:book:](.\u002Fcitations\u002Floo2022efficient.txt)\n+ [Dataset Distillation with Convexified Implicit Gradients](https:\u002F\u002Farxiv.org\u002Fabs\u002F2302.06755) (Noel Loo et al., ICML 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Fyolky\u002FRCIG) [:book:](.\u002Fcitations\u002Floo2023dataset.txt)\n+ [Provable and Efficient Dataset Distillation for Kernel Ridge Regression](https:\u002F\u002Fopenreview.net\u002Fforum?id=WI2VpcBdnd) (Yilan Chen et al., NeurIPS 2024) [:book:](.\u002Fcitations\u002Fchen2024krr.txt)\n\n\u003Ca name=\"parametrization\" \u002F>\n\n### Distilled Dataset Parametrization\n+ [Dataset Condensation via Efficient Synthetic-Data Parameterization](https:\u002F\u002Farxiv.org\u002Fabs\u002F2205.14959) (Jang-Hyun Kim et al., ICML 2022) [:octocat:](https:\u002F\u002Fgithub.com\u002Fsnu-mllab\u002Fefficient-dataset-condensation) [:book:](.\u002Fcitations\u002Fkim2022dataset.txt)\n+ [Remember the Past: Distilling Datasets into Addressable Memories for Neural Networks](https:\u002F\u002Farxiv.org\u002Fabs\u002F2206.02916) (Zhiwei Deng et al., NeurIPS 2022) [:octocat:](https:\u002F\u002Fgithub.com\u002Fprincetonvisualai\u002FRememberThePast-DatasetDistillation) [:book:](.\u002Fcitations\u002Fdeng2022remember.txt)\n+ [On Divergence Measures for Bayesian Pseudocoresets](https:\u002F\u002Farxiv.org\u002Fabs\u002F2210.06205) (Balhae Kim et al., NeurIPS 2022) [:octocat:](https:\u002F\u002Fgithub.com\u002Fbalhaekim\u002Fbpc-divergences) [:book:](.\u002Fcitations\u002Fkim2022divergence.txt)\n+ [Dataset Distillation via Factorization](https:\u002F\u002Farxiv.org\u002Fabs\u002F2210.16774) (Songhua Liu et al., NeurIPS 2022) [:octocat:](https:\u002F\u002Fgithub.com\u002FHuage001\u002FDatasetFactorization) [:book:](.\u002Fcitations\u002Fliu2022dataset.txt)\n+ [PRANC: Pseudo RAndom Networks for Compacting Deep Models](https:\u002F\u002Farxiv.org\u002Fabs\u002F2206.08464) (Parsa Nooralinejad et al., 2022) [:octocat:](https:\u002F\u002Fgithub.com\u002FUCDvision\u002FPRANC) [:book:](.\u002Fcitations\u002Fnooralinejad2022pranc.txt)\n+ [Dataset Condensation with Latent Space Knowledge Factorization and Sharing](https:\u002F\u002Farxiv.org\u002Fabs\u002F2208.10494) (Hae Beom Lee & Dong Bok Lee et al., 2022) [:book:](.\u002Fcitations\u002Flee2022kfs.txt)\n+ [Slimmable Dataset Condensation](https:\u002F\u002Fopenaccess.thecvf.com\u002Fcontent\u002FCVPR2023\u002Fhtml\u002FLiu_Slimmable_Dataset_Condensation_CVPR_2023_paper.html) (Songhua Liu et al., CVPR 2023) [:book:](.\u002Fcitations\u002Fliu2023slimmable.txt)\n+ [Few-Shot Dataset Distillation via Translative Pre-Training](https:\u002F\u002Fopenaccess.thecvf.com\u002Fcontent\u002FICCV2023\u002Fhtml\u002FLiu_Few-Shot_Dataset_Distillation_via_Translative_Pre-Training_ICCV_2023_paper.html) (Songhua Liu et al., ICCV 2023) [:book:](.\u002Fcitations\u002Fliu2023fewshot.txt)\n+ [MGDD: A Meta Generator for Fast Dataset Distillation](https:\u002F\u002Fopenreview.net\u002Fforum?id=D9CMRR5Lof) (Songhua Liu et al., NeurIPS 2023) [:book:](.\u002Fcitations\u002Fliu2023mgdd.txt)\n+ [Sparse Parameterization for Epitomic Dataset Distillation](https:\u002F\u002Fopenreview.net\u002Fforum?id=ZIfhYAE2xg) (Xing Wei & Anjia Cao et al., NeurIPS 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002FMIV-XJTU\u002FSPEED) [:book:](.\u002Fcitations\u002Fwei2023sparse.txt)\n+ [Frequency Domain-based Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2311.08819) (Donghyeok Shin & Seungjae Shin et al., NeurIPS 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Fsdh0818\u002FFreD) [:book:](.\u002Fcitations\u002Fshin2023fred.txt)\n+ [Leveraging Hierarchical Feature Sharing for Efficient Dataset Condensation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.07506) (Haizhong Zheng et al., ECCV 2024) [:book:](.\u002Fcitations\u002Fzheng2024hmn.txt)\n+ [FYI: Flip Your Images for Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2407.08113) (Byunggwan Son et al., ECCV 2024) [:globe_with_meridians:](https:\u002F\u002Fcvlab.yonsei.ac.kr\u002Fprojects\u002FFYI\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fcvlab-yonsei\u002FFYI) [:book:](.\u002Fcitations\u002Fson2024fyi.txt)\n+ [Color-Oriented Redundancy Reduction in Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2411.11329) (Bowen Yuan et al., NeurIPS 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FKeViNYuAn0314\u002FAutoPalette) [:book:](.\u002Fcitations\u002Fyuan2024color.txt)\n+ [Distilling Dataset into Neural Field](https:\u002F\u002Farxiv.org\u002Fabs\u002F2503.04835) (Donghyeok Shin et al., ICLR 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002Faailab-kaist\u002FDDiF) [:book:](.\u002Fcitations\u002Fshin2025ddif.txt)\n+ [Dataset Distillation as Data Compression: A Rate-Utility Perspective](https:\u002F\u002Farxiv.org\u002Fabs\u002F2507.17221) (Youneng Bao & Yiping Liu et al., ICCV 2025) [:globe_with_meridians:](https:\u002F\u002Fnouise.github.io\u002FDD-RUO\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fnouise\u002FDD-RUO) [:book:](.\u002Fcitations\u002Fbao2025ruo.txt)\n+ [Beyond Pixels: Efficient Dataset Distillation via Sparse Gaussian Representation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2509.26219) (Chenyang Jiang et al., 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002Fj-cyoung\u002FGSDatasetDistillation) [:book:](.\u002Fcitations\u002Fjiang2025gsdd.txt)\n+ [Post Training Quantization for Efficient Dataset Condensation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.13346) (Linh-Tam Tran et al., AAAI 2026) [:book:](.\u002Fcitations\u002Ftran2026ptqdc.txt)\n\n\u003Ca name=\"generative\" \u002F>\n\n### Generative Distillation\n#### GAN\n+ [Synthesizing Informative Training Samples with GAN](https:\u002F\u002Farxiv.org\u002Fabs\u002F2204.07513) (Bo Zhao et al., NeurIPS 2022 Workshop) [:octocat:](https:\u002F\u002Fgithub.com\u002Fvico-uoe\u002Fit-gan) [:book:](.\u002Fcitations\u002Fzhao2022synthesizing.txt)\n+ [Generalizing Dataset Distillation via Deep Generative Prior](https:\u002F\u002Farxiv.org\u002Fabs\u002F2305.01649) (George Cazenavette et al., CVPR 2023) [:globe_with_meridians:](https:\u002F\u002Fgeorgecazenavette.github.io\u002Fglad\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fgeorgecazenavette\u002Fglad) [:book:](.\u002Fcitations\u002Fcazenavette2023glad.txt)\n+ [DiM: Distilling Dataset into Generative Model](https:\u002F\u002Farxiv.org\u002Fabs\u002F2303.04707) (Kai Wang & Jianyang Gu et al., 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Fvimar-gu\u002FDiM) [:book:](.\u002Fcitations\u002Fwang2023dim.txt)\n+ [Dataset Condensation via Generative Model](https:\u002F\u002Farxiv.org\u002Fabs\u002F2309.07698) (Junhao Zhang et al., 2023) [:book:](.\u002Fcitations\u002Fzhang2023dc.txt)\n+ [Generative Dataset Distillation: Balancing Global Structure and Local Details](https:\u002F\u002Farxiv.org\u002Fabs\u002F2404.17732) (Longzhen Li & Guang Li et al., CVPR 2024 Workshop) [:book:](.\u002Fcitations\u002Fli2024generative.txt)\n+ [Data-to-Model Distillation: Data-Efficient Learning Framework](https:\u002F\u002Fwww.ecva.net\u002Fpapers\u002Feccv_2024\u002Fpapers_ECCV\u002Fhtml\u002F6020_ECCV_2024_paper.php) (Ahmad Sajedi & Samir Khaki et al., ECCV 2024) [:book:](.\u002Fcitations\u002Fsajedi2024data.txt)\n+ [Generative Dataset Distillation Based on Self-knowledge Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2501.04202) (Longzhen Li & Guang Li et al., ICASSP 2025) [:book:](.\u002Fcitations\u002Fli2025generative.txt)\n+ [Hierarchical Features Matter: A Deep Exploration of GAN Priors for Improved Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.05704) (Xinhao Zhong & Hao Fang et al., CVPR 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002Fndhg1213\u002FH-GLaD) [:book:](.\u002Fcitations\u002Fzhong2025hglad.txt)\n\n#### Diffusion\n+ [Efficient Dataset Distillation via Minimax Diffusion](https:\u002F\u002Farxiv.org\u002Fabs\u002F2311.15529) (Jianyang Gu et al., CVPR 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fvimar-gu\u002FMinimaxDiffusion) [:book:](.\u002Fcitations\u002Fgu2024efficient.txt)\n+ [D4M: Dataset Distillation via Disentangled Diffusion Model](https:\u002F\u002Farxiv.org\u002Fabs\u002F2407.15138) (Duo Su & Junjie Hou et al., CVPR 2024) [:globe_with_meridians:](https:\u002F\u002Fjunjie31.github.io\u002FD4M\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fsuduo94\u002FD4M) [:book:](.\u002Fcitations\u002Fsu2024d4m.txt)\n+ [Generative Dataset Distillation Based on Diffusion Model](https:\u002F\u002Farxiv.org\u002Fabs\u002F2408.08610) (Duo Su & Junjie Hou & Guang Li et al., ECCV 2024 Workshop) [:octocat:](https:\u002F\u002Fgithub.com\u002FGuang000\u002FGenerative-Dataset-Distillation-Based-on-Diffusion-Model) [:book:](.\u002Fcitations\u002Fsu2024diffusion.txt)\n+ [Influence-Guided Diffusion for Dataset Distillation](https:\u002F\u002Fopenreview.net\u002Fforum?id=0whx8MhysK) (Mingyang Chen et al., ICLR 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002Fmchen725\u002FDD_IGD) [:book:](.\u002Fcitations\u002Fchen2025igd.txt)\n+ [Taming Diffusion for Dataset Distillation with High Representativeness](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.18399) (Lin Zhao et al., ICML 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002Flin-zhao-resoLve\u002FD3HR) [:book:](.\u002Fcitations\u002Fzhao2025d3hr.txt)\n+ [MGD3: Mode-Guided Dataset Distillation using Diffusion Models](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.18963) (Jeffrey A. Chan-Santiago et al., ICML 2025) [:globe_with_meridians:](https:\u002F\u002Fjachansantiago.com\u002Fmode-guided-distillation\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fjachansantiago\u002Fmode_guidance\u002F) [:book:](.\u002Fcitations\u002Fchan-santiago2025mgd3.txt)\n+ [Enhancing Diffusion-based Dataset Distillation via Adversary-Guided Curriculum Sampling](https:\u002F\u002Farxiv.org\u002Fabs\u002F2508.01264) (Lexiao Zou et al., ICME 2025) [:book:](.\u002Fcitations\u002Fzou2025acs.txt)\n+ [CaO2: Rectifying Inconsistencies in Diffusion-Based Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2506.22637) (Haoxuan Wang et al., ICCV 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002FhatchetProject\u002FCaO2) [:book:](.\u002Fcitations\u002Fwang2025cao2.txt)\n+ [Dataset Distillation via Vision-Language Category Prototype](https:\u002F\u002Farxiv.org\u002Fabs\u002F2506.23580) (Yawen Zou & Guang Li et al., ICCV 2025) [:globe_with_meridians:](https:\u002F\u002Fzou-yawen.github.io\u002FDD_via_vision-language)  [:octocat:](https:\u002F\u002Fgithub.com\u002Fzou-yawen\u002FDataset-Distillation-via-Vision-Language-Category-Prototype\u002F) [:book:](.\u002Fcitations\u002Fzou2025vlcp.txt)\n+ [Task-Specific Generative Dataset Distillation with Difficulty-Guided Sampling](https:\u002F\u002Farxiv.org\u002Fabs\u002F2507.03331) (Mingzhuo Li & Guang Li et al., ICCV 2025 Workshop) [:octocat:](https:\u002F\u002Fgithub.com\u002FSumomoTaku\u002FDiffGuideSamp) [:book:](.\u002Fcitations\u002Fli2025diff.txt)\n+ [Unlocking Dataset Distillation with Diffusion Models](https:\u002F\u002Farxiv.org\u002Fabs\u002F2403.03881) (Brian B. Moser & Federico Raue et al., NeurIPS 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002FBrian-Moser\u002Fprune_and_distill) [:book:](.\u002Fcitations\u002Fmoser2025ld3m.txt)\n+ [Optimizing Distributional Geometry Alignment with Optimal Transport for Generative Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2512.00308) (Xiao Cui et al., NeurIPS 2025) [:book:](.\u002Fcitations\u002Fcui2025ot.txt)\n+ [Dataset Condensation with Color Compensation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2508.01139) (Huyu Wu et al., TMLR 2025) [:globe_with_meridians:](https:\u002F\u002F528why.github.io\u002FDC3-Page\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002F528why\u002FDataset-Condensation-with-Color-Compensation) [:book:](.\u002Fcitations\u002Fwu2025dc3.txt)\n+ [Diffusion Models as Dataset Distillation Priors](https:\u002F\u002Farxiv.org\u002Fabs\u002F2510.17421) (Duo Su et al., ICLR 2026) [:book:](.\u002Fcitations\u002Fsu2026dap.txt)\n+ [CoDA: From Text-to-Image Diffusion Models to Training-Free Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2512.03844) (Letian Zhou et al., ICLR 2026) [:octocat:](https:\u002F\u002Fgithub.com\u002Fzzzlt422\u002FCoDA) [:book:](.\u002Fcitations\u002Fzhou2026coda.txt)\n+ [ManifoldGD: Training-Free Hierarchical Manifold Guidance for Diffusion-Based Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2602.23295) (Ayush Roy et al., CVPR 2026) [:octocat:](https:\u002F\u002Fgithub.com\u002FAyushRoy2001\u002FManifoldGD) [:book:](.\u002Fcitations\u002Froy2026manifold.txt)\n+ [IMS3: Breaking Distributional Aggregation in Diffusion-Based Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.13960) (Chenru Wang & Yunyi Chen et al., CVPR 2026) [:book:](.\u002Fcitations\u002Fwang2026ims3.txt)\n+ [EVLF: Early Vision-Language Fusion for Generative Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.07476) (Wenqi Cai et al., CVPR 2026) [:globe_with_meridians:](https:\u002F\u002Fwenqi-cai297.github.io\u002Fearlyfusion-HP\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fwenqi-cai297\u002Fearlyfusion-for-dd\u002F) [:book:](.\u002Fcitations\u002Fcai2026evlf.txt)\n+ [Learnability-Guided Diffusion for Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2604.00519) (Jeffrey A. Chan-Santiago et al., CVPR 2026) [:globe_with_meridians:](https:\u002F\u002Fjachansantiago.com\u002Flearnability-guided-distillation\u002F) [:book:](.\u002Fcitations\u002Fchansantiago2026learnability.txt)\n\n#### VAR\n+ [HIERAMP: Coarse-to-Fine Autoregressive Amplification for Generative Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.06932) (Lin Zhao & Xinru Jiang et al., CVPR 2026) [:octocat:](https:\u002F\u002Fgithub.com\u002FOshikaka\u002FHIERAMP) [:book:](.\u002Fcitations\u002Fzhao2026hieramp.txt)\n\n#### Flow\n+ [Path-Guided Flow Matching for Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2602.05616) (Xuhui Li et al., 2026) [:book:](.\u002Fcitations\u002Fli2026flow.txt)\n\n\u003Ca name=\"optimization\" \u002F>\n\n### Better Optimization\n+ [Accelerating Dataset Distillation via Model Augmentation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2212.06152) (Lei Zhang & Jie Zhang et al., CVPR 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Fncsu-dk-lab\u002FAcc-DD) [:book:](.\u002Fcitations\u002Fzhang2023accelerating.txt)\n+ [DREAM: Efficient Dataset Distillation by Representative Matching](https:\u002F\u002Farxiv.org\u002Fabs\u002F2302.14416) (Yanqing Liu & Jianyang Gu & Kai Wang et al., ICCV 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Flyq312318224\u002FDREAM) [:book:](.\u002Fcitations\u002Fliu2023dream.txt)\n+ [You Only Condense Once: Two Rules for Pruning Condensed Datasets](https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.14019) (Yang He et al., NeurIPS 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Fhe-y\u002Fyou-only-condense-once) [:book:](.\u002Fcitations\u002Fhe2023yoco.txt)\n+ [MIM4DD: Mutual Information Maximization for Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2312.16627) (Yuzhang Shang et al., NeurIPS 2023) [:book:](.\u002Fcitations\u002Fshang2023mim4dd.txt)\n+ [Can Pre-Trained Models Assist in Dataset Distillation?](https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.03295) (Yao Lu et al., 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Fyaolu-zjut\u002FDDInterpreter) [:book:](.\u002Fcitations\u002Flu2023pre.txt)\n+ [DREAM+: Efficient Dataset Distillation by Bidirectional Representative Matching](https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.15052) (Yanqing Liu & Jianyang Gu & Kai Wang et al., 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Flyq312318224\u002FDREAM) [:book:](.\u002Fcitations\u002Fliu2023dream+.txt)\n+ [Dataset Distillation in Latent Space](https:\u002F\u002Farxiv.org\u002Fabs\u002F2311.15547) (Yuxuan Duan et al., 2023) [:book:](.\u002Fcitations\u002Fduan2023latent.txt)\n+ [Data Distillation Can Be Like Vodka: Distilling More Times For Better Quality](https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.06982) (Xuxi Chen & Yu Yang et al., ICLR 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FVITA-Group\u002FProgressiveDD) [:book:](.\u002Fcitations\u002Fchen2024vodka.txt)\n+ [Embarassingly Simple Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2311.07025) (Yunzhen Feng et al., ICLR 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Ffengyzpku\u002FSimple_Dataset_Distillation) [:book:](.\u002Fcitations\u002Fyunzhen2024embarassingly.txt)\n+ [Multisize Dataset Condensation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2403.06075) (Yang He et al., ICLR 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fhe-y\u002FMultisize-Dataset-Condensation) [:book:](.\u002Fcitations\u002Fhe2024mdc.txt)\n+ [Large Scale Dataset Distillation with Domain Shift](https:\u002F\u002Fopenreview.net\u002Fforum?id=0FWPKHMCSc) (Noel Loo & Alaa Maalouf et al., ICML 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fyolky\u002Fd3s_distillation) [:book:](.\u002Fcitations\u002Floo2024d3s.txt)\n+ [Distill Gold from Massive Ores: Bi-level Data Pruning towards Efficient Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2305.18381) (Yue Xu et al., ECCV 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fsilicx\u002FGoldFromOres) [:book:](.\u002Fcitations\u002Fxu2024distill.txt)\n+ [Towards Model-Agnostic Dataset Condensation by Heterogeneous Models](https:\u002F\u002Farxiv.org\u002Fabs\u002F2409.14538) (Jun-Yeong Moon et al., ECCV 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fkhu-agi\u002Fhmdc) [:book:](.\u002Fcitations\u002Fmoon2024hmdc.txt)\n+ [Teddy: Efficient Large-Scale Dataset Distillation via Taylor-Approximated Matching](https:\u002F\u002Farxiv.org\u002Fabs\u002F2410.07579) (Ruonan Yu et al., ECCV 2024) [:book:](.\u002Fcitations\u002Fyu2024teddy.txt)\n+ [BACON: Bayesian Optimal Condensation Framework for Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.01112) (Zheng Zhou et al., 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fzhouzhengqd\u002FBACON) [:book:](.\u002Fcitations\u002Fzhou2024bacon.txt)\n+ [Going Beyond Feature Similarity: Effective Dataset Distillation based on Class-aware Conditional Mutual Information](https:\u002F\u002Farxiv.org\u002Fabs\u002F2412.09945) (Xinhao Zhong et al., ICLR 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002Fndhg1213\u002FCMIDD) [:book:](.\u002Fcitations\u002Fzhong2025cmi.txt)\n+ [Curriculum Coarse-to-Fine Selection for High-IPC Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2503.18872) (Yanda Chen & Gongwei Chen et al., CVPR 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002FCYDaaa30\u002FCCFS) [:book:](.\u002Fcitations\u002Fchen2025ccfs.txt)\n+ [Not All Samples Should Be Utilized Equally: Towards Understanding and Improving Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2408.12483) (Shaobo Wang et al., CVPR 2025 Workshop) [:book:](.\u002Fcitations\u002Fwang2025samples.txt)\n+ [Beyond Random: Automatic Inner-loop Optimization in Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2510.04838) (Muquan Li et al., NeurIPS 2025) [:book:](.\u002Fcitations\u002Fli2025bptt.txt)\n+ [Dataset Distillation as Pushforward Optimal Quantization](https:\u002F\u002Farxiv.org\u002Fabs\u002F2501.07681) (Hongye Tan et al., ICLR 2026) [:book:](.\u002Fcitations\u002Ftan2026optimal.txt)\n\n\n\u003Ca name=\"understanding\" \u002F>\n\n### Better Understanding\n+ [Optimizing Millions of Hyperparameters by Implicit Differentiation](https:\u002F\u002Farxiv.org\u002Fabs\u002F1911.02590) (Jonathan Lorraine et al., AISTATS 2020) [:octocat:](https:\u002F\u002Fgithub.com\u002FMaximeVandegar\u002FPapers-in-100-Lines-of-Code\u002Ftree\u002Fmain\u002FOptimizing_Millions_of_Hyperparameters_by_Implicit_Differentiation) [:book:](.\u002Fcitations\u002Florraine2020optimizing.txt) \n+ [On Implicit Bias in Overparameterized Bilevel Optimization](https:\u002F\u002Fproceedings.mlr.press\u002Fv162\u002Fvicol22a.html) (Paul Vicol et al., ICML 2022) [:book:](.\u002Fcitations\u002Fvicol2022implicit.txt)\n+ [On the Size and Approximation Error of Distilled Sets](https:\u002F\u002Farxiv.org\u002Fabs\u002F2305.14113) (Alaa Maalouf & Murad Tukan et al., NeurIPS 2023) [:book:](.\u002Fcitations\u002Fmaalouf2023size.txt)\n+ [A Theoretical Study of Dataset Distillation](https:\u002F\u002Fopenreview.net\u002Fforum?id=dq5QGXGxoJ) (Zachary Izzo et al., NeurIPS 2023 Workshop) [:book:](.\u002Fcitations\u002Fizzo2023theo.txt)\n+ [What is Dataset Distillation Learning?](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.04284) (William Yang et al., ICML 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fprincetonvisualai\u002FWhat-is-Dataset-Distillation-Learning) [:book:](.\u002Fcitations\u002Fyang2024learning.txt)\n+ [Mitigating Bias in Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.06609) (Justin Cui et al., ICML 2024) [:book:](.\u002Fcitations\u002Fcui2024bias.txt)\n+ [Dataset Distillation from First Principles: Integrating Core Information Extraction and Purposeful Learning](https:\u002F\u002Farxiv.org\u002Fabs\u002F2409.01410) (Vyacheslav Kungurtsev et al., 2024) [:book:](.\u002Fcitations\u002Fkungurtsev2024first.txt)\n+ [Information-Guided Diffusion Sampling for Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2507.04619) (Linfeng Ye et al., NeurIPS 2025 Workshop) [:book:](.\u002Fcitations\u002Fye2025igds.txt)\n+ [A Discrepancy-Based Perspective on Dataset Condensation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2509.10367) (Tong Chen et al., 2025) [:book:](.\u002Fcitations\u002Fchen2025discrepancy.txt)\n+ [Understanding Dataset Distillation via Spectral Filtering](https:\u002F\u002Farxiv.org\u002Fabs\u002F2503.01212) (Deyu Bo et al., ICLR 2026) [:book:](.\u002Fcitations\u002Fbo2026unidd.txt)\n+ [Dataset Distillation for Memorized Data: Soft Labels can Leak Held-Out Teacher Knowledge](https:\u002F\u002Farxiv.org\u002Fabs\u002F2506.14457) (Freya Behrens et al., ICLR 2026) [:book:](.\u002Fcitations\u002Fbehrens2026soft.txt)\n+ [Dataset Distillation Efficiently Encodes Low-Dimensional Representations from Gradient-Based Learning of Non-Linear Tasks](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.14830) (Yuri Kinoshita et al., 2026) [:book:](.\u002Fcitations\u002Fkinoshita2026lowdim.txt)\n\n\u003Ca name=\"label\" \u002F>\n\n### Label Distillation\n+ [Flexible Dataset Distillation: Learn Labels Instead of Images](https:\u002F\u002Farxiv.org\u002Fabs\u002F2006.08572) (Ondrej Bohdal et al., NeurIPS 2020 Workshop) [:octocat:](https:\u002F\u002Fgithub.com\u002Fondrejbohdal\u002Flabel-distillation) [:book:](.\u002Fcitations\u002Fbohdal2020flexible.txt)\n+ [Soft-Label Dataset Distillation and Text Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F1910.02551) (Ilia Sucholutsky et al., IJCNN 2021) [:octocat:](https:\u002F\u002Fgithub.com\u002Filia10000\u002Fdataset-distillation) [:book:](.\u002Fcitations\u002Fsucholutsky2021soft.txt)\n+ [A Label is Worth a Thousand Images in Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.10485) (Tian Qin et al., NeurIPS 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fsunnytqin\u002Fno-distillation) [:book:](.\u002Fcitations\u002Fqin2024label.txt)\n+ [Are Large-scale Soft Labels Necessary for Large-scale Dataset Distillation?](https:\u002F\u002Farxiv.org\u002Fabs\u002F2410.15919) (Lingao Xiao et al., NeurIPS 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fhe-y\u002Fsoft-label-pruning-for-dataset-distillation) [:book:](.\u002Fcitations\u002Fxiao2024soft.txt)\n+ [DRUPI: Dataset Reduction Using Privileged Information](https:\u002F\u002Farxiv.org\u002Fabs\u002F2410.01611) (Shaobo Wang et al., 2024) [:book:](.\u002Fcitations\u002Fwang2024drupi.txt)\n+ [Label-Augmented Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2409.16239) (Seoungyoon Kang & Youngsun Lim et al., WACV 2025) [:book:](.\u002Fcitations\u002Fkang2024label.txt)\n+ [GIFT: Unlocking Full Potential of Labels in Distilled Dataset at Near-zero Cost](https:\u002F\u002Farxiv.org\u002Fabs\u002F2405.14736) (Xinyi Shang & Peng Sun et al., ICLR 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002FLINs-lab\u002FGIFT) [:book:](.\u002Fcitations\u002Fshang2025gift.txt)\n+ [Heavy Labels Out! Dataset Distillation with Label Space Lightening](https:\u002F\u002Farxiv.org\u002Fabs\u002F2408.08201) (Ruonan Yu et al., ICCV 2025) [:book:](.\u002Fcitations\u002Fyu2025helio.txt)\n\n\u003Ca name=\"quant\" \u002F>\n\n### Dataset Quantization\n+ [Dataset Quantization](https:\u002F\u002Farxiv.org\u002Fabs\u002F2308.10524) (Daquan Zhou & Kai Wang & Jianyang Gu et al., ICCV 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Fmagic-research\u002FDataset_Quantization) [:book:](.\u002Fcitations\u002Fzhou2023dataset.txt)\n+ [Dataset Quantization with Active Learning based Adaptive Sampling](https:\u002F\u002Farxiv.org\u002Fabs\u002F2407.07268) (Zhenghao Zhao et al., ECCV 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fichbill\u002FDQAS) [:book:](.\u002Fcitations\u002Fzhao2024dqas.txt)\n+ [Adaptive Dataset Quantization](https:\u002F\u002Fwww.arxiv.org\u002Fabs\u002F2412.16895) (Muquan Li et al., AAAI 2025) [:book:](.\u002Fcitations\u002Fli2025adq.txt)\n+ [Dataset Color Quantization: A Training-Oriented Framework for Dataset-Level Compression](https:\u002F\u002Farxiv.org\u002Fabs\u002F2602.20650) (Chenyue Yu et al., ICLR 2026) [:book:](.\u002Fcitations\u002Fyu2026dcq.txt)\n\n\u003Ca name=\"decouple\" \u002F>\n\n### Decoupled Distillation\n+ [Squeeze, Recover and Relabel: Dataset Condensation at ImageNet Scale From A New Perspective](https:\u002F\u002Farxiv.org\u002Fabs\u002F2306.13092) (Zeyuan Yin & Zhiqiang Shen et al., NeurIPS 2023) [:globe_with_meridians:](https:\u002F\u002Fzeyuanyin.github.io\u002Fprojects\u002FSRe2L\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002FVILA-Lab\u002FSRe2L\u002Ftree\u002Fmain\u002FSRe2L) [:book:](.\u002Fcitations\u002Fyin2023sre2l.txt)\n+ [Dataset Distillation via Curriculum Data Synthesis in Large Data Era](https:\u002F\u002Farxiv.org\u002Fabs\u002F2311.18838) (Zeyuan Yin et al., TMLR 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FVILA-Lab\u002FSRe2L\u002Ftree\u002Fmain\u002FCDA) [:book:](.\u002Fcitations\u002Fyin2024cda.txt)\n+ [Generalized Large-Scale Data Condensation via Various Backbone and Statistical Matching](https:\u002F\u002Farxiv.org\u002Fabs\u002F2311.17950) (Shitong Shao et al., CVPR 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fshaoshitong\u002FG_VBSM_Dataset_Condensation) [:book:](.\u002Fcitations\u002Fshao2024gvbsm.txt)\n+ [On the Diversity and Realism of Distilled Dataset: An Efficient Dataset Distillation Paradigm](https:\u002F\u002Farxiv.org\u002Fabs\u002F2312.03526) (Peng Sun et al., CVPR 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FLINs-lab\u002FRDED) [:book:](.\u002Fcitations\u002Fsun2024rded.txt)\n+ [Information Compensation: A Fix for Any-scale Dataset Distillation](https:\u002F\u002Fopenreview.net\u002Fforum?id=2SnmKd1JK4) (Peng Sun et al., ICLR 2024 Workshop) [:book:](.\u002Fcitations\u002Fsun2024lic.txt)\n+ [Elucidating the Design Space of Dataset Condensation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2404.13733) (Shitong Shao et al., NeurIPS 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fshaoshitong\u002FEDC) [:book:](.\u002Fcitations\u002Fshao2024edc.txt)\n+ [Diversity-Driven Synthesis: Enhancing Dataset Distillation through Directed Weight Adjustment](https:\u002F\u002Farxiv.org\u002Fabs\u002F2409.17612) (Jiawei Du et al., NeurIPS 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FAngusDujw\u002FDiversity-Driven-Synthesis) [:book:](.\u002Fcitations\u002Fdu2024diversity.txt)\n+ [Breaking Class Barriers: Efficient Dataset Distillation via Inter-Class Feature Compensator](https:\u002F\u002Farxiv.org\u002Fabs\u002F2408.06927) (Xin Zhang et al., ICLR 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002Fzhangxin-xd\u002FUFC) [:book:](.\u002Fcitations\u002Fzhang2025infer.txt)\n+ [DELT: A Simple Diversity-driven EarlyLate Training for Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2411.19946) (Zhiqiang Shen & Ammar Sherif et al., CVPR 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002FVILA-Lab\u002FDELT) [:book:](.\u002Fcitations\u002Fshen2025delt.txt)\n+ [Enhancing Dataset Distillation via Non-Critical Region Refinement](https:\u002F\u002Farxiv.org\u002Fabs\u002F2503.18267) (Minh-Tuan Tran et al., CVPR 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002Ftmtuan1307\u002FNRR-DD) [:book:](.\u002Fcitations\u002Ftran2025nrrdd.txt)\n+ [Curriculum Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2405.09150) (Zhiheng Ma & Anjia Cao et al., TIP 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002FMIV-XJTU\u002FCUDD) [:book:](.\u002Fcitations\u002Fma2025cudd.txt)\n+ [FADRM: Fast and Accurate Data Residual Matching for Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2506.24125) (Jiacheng Cui & Xinyue Bi et al., NeurIPS 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002FJiacheng8\u002FFADRM) [:book:](.\u002Fcitations\u002Fcui2025fadrm.txt)\n+ [FocusDD: Real-World Scene Infusion for Robust Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2501.06405) (Youbin Hu et al., 2025) [:book:](.\u002Fcitations\u002Fhu2025focusdd.txt)\n+ [Dataset Distillation via Committee Voting](https:\u002F\u002Farxiv.org\u002Fabs\u002F2501.07575) (Jiacheng Cui et al., 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002FJiacheng8\u002FCV-DD) [:book:](.\u002Fcitations\u002Fcui2025cvdd.txt)\n+ [PRISM: Diversifying Dataset Distillation by Decoupling Architectural Priors](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.09905) (Brian B. Moser et al., TMLR 2026) [:octocat:](https:\u002F\u002Fgithub.com\u002FBrian-Moser\u002Fprism) [:book:](.\u002Fcitations\u002Fmoser2026prism.txt)\n+ [DiRe: Diversity-promoting Regularization for Dataset Condensation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2512.13083) (Saumyaranjan Mohanty et al., WACV 2026) [:octocat:](https:\u002F\u002Fgithub.com\u002FDIL-IITH\u002FDiRe) [:book:](.\u002Fcitations\u002Fmohanty2026dire.txt)\n+ [Grounding and Enhancing Informativeness and Utility in Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2601.21296) (Shaobo Wang et al., ICLR 2026) [:book:](.\u002Fcitations\u002Fwang2026infoutil.txt)\n+ [Fixed Anchors Are Not Enough: Dynamic Retrieval and Persistent Homology for Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2602.24144) (Muquan Li et al., CVPR 2026) [:book:](.\u002Fcitations\u002Fli2026reta.txt)\n\n\u003Ca name=\"multi\" \u002F>\n\n### Multimodal Distillation\n+ [Vision-Language Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2308.07545) (Xindi Wu et al., TMLR 2024) [:globe_with_meridians:](https:\u002F\u002Fprincetonvisualai.github.io\u002Fmultimodal_dataset_distillation\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fprincetonvisualai\u002Fmultimodal_dataset_distillation) [:book:](.\u002Fcitations\u002Fwu2024multi.txt)\n+ [Low-Rank Similarity Mining for Multimodal Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.03793) (Yue Xu et al., ICML 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fsilicx\u002FLoRS_Distill) [:book:](.\u002Fcitations\u002Fxu2024lors.txt)\n+ [Audio-Visual Dataset Distillation](https:\u002F\u002Fopenreview.net\u002Fforum?id=IJlbuSrXmk) (Saksham Singh Kushwaha et al., TMLR 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fsakshamsingh1\u002FAVDD) [:book:](.\u002Fcitations\u002Fkush2024avdd.txt)\n+ [Beyond Modality Collapse: Representations Blending for Multimodal Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.14705) (Xin Zhang et al., NeurIPS 2025) [:book:](.\u002Fcitations\u002Fzhang2025mdd.txt)\n+ [Efficient Multimodal Dataset Distillation via Generative Models](https:\u002F\u002Farxiv.org\u002Fabs\u002F2509.15472) (Zhenghao Zhao et al., NeurIPS 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002Fichbill\u002FEDGE) [:book:](.\u002Fcitations\u002Fzhao2025edge.txt)\n+ [CovMatch: Cross-Covariance Guided Multimodal Dataset Distillation with Trainable Text Encoder](https:\u002F\u002Farxiv.org\u002Fabs\u002F2510.18583) (Yongmin Lee et al., NeurIPS 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002FYongalls\u002FCovMatch) [:book:](.\u002Fcitations\u002Flee2025covmatch.txt)\n+ [Decoupled Audio-Visual Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.17890) (Wenyuan Li & Guang Li et al., 2025) [:book:](.\u002Fcitations\u002Fli2025davdd.txt)\n+ [ImageBindDC: Compressing Multi-modal Data with ImageBind-based Condensation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.08263) (Yue Min & Shaobo Wang et al., AAAI 2026) [:book:](.\u002Fcitations\u002Fmin2026imagebinddc.txt)\n+ [Multimodal Dataset Distillation Made Simple by Prototype-Guided Data Synthesis](https:\u002F\u002Farxiv.org\u002Fabs\u002F2602.19756) (Junhyeok Choi et al., ICLR 2026) [:octocat:](https:\u002F\u002Fgithub.com\u002Fjunhyeok9712\u002FPDS) [:book:](.\u002Fcitations\u002Fchoi2026multi.txt)\n+ [Multimodal Dataset Distillation via Phased Teacher Models](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.25388) (Shengbin Guo & Hang Zhao et al., ICLR 2026) [:octocat:](https:\u002F\u002Fgithub.com\u002FPrevisior\u002FPTM-ST) [:book:](.\u002Fcitations\u002Fguo2026ptmst.txt)\n+ [Asynchronous Matching with Dynamic Sampling for Multimodal Dataset Distillation](https:\u002F\u002Fopenreview.net\u002Fforum?id=7SgSMKM2KF) (Ding Qi et al., ICLR 2026) [:book:](.\u002Fcitations\u002Fqi2026amd.txt)\n\n\u003Ca name=\"self\" \u002F>\n\n### Self-Supervised Distillation\n+ [Self-Supervised Dataset Distillation for Transfer Learning](https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.06511) (Dong Bok Lee & Seanie Lee et al., ICLR 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fdb-Lee\u002Fselfsup_dd) [:book:](.\u002Fcitations\u002Flee2024self.txt)\n+ [Efficiency for Free: Ideal Data Are Transportable Representations](https:\u002F\u002Farxiv.org\u002Fabs\u002F2405.14669) (Peng Sun et al., NeurIPS 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FLINs-lab\u002FReLA) [:book:](.\u002Fcitations\u002Fsun2024rela.txt)\n+ [Self-supervised Dataset Distillation: A Good Compression Is All You Need](https:\u002F\u002Farxiv.org\u002Fabs\u002F2404.07976) (Muxin Zhou et al., 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FVILA-Lab\u002FSRe2L\u002Ftree\u002Fmain\u002FSCDD\u002F) [:book:](.\u002Fcitations\u002Fzhou2024self.txt)\n+ [Dataset Distillation via Knowledge Distillation: Towards Efficient Self-Supervised Pre-Training of Deep Networks](https:\u002F\u002Farxiv.org\u002Fabs\u002F2410.02116) (Siddharth Joshi et al., ICLR 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002Fjiayini1119\u002FMKDT) [:book:](.\u002Fcitations\u002Fjoshi2025kd.txt)\n+ [Boost Self-Supervised Dataset Distillation via Parameterization, Predefined Augmentation, and Approximation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2507.21455) (Sheng-Feng Yu et al., ICLR 2025) [:book:](.\u002Fcitations\u002Fyu2025self.txt)\n+ [Dataset Distillation for Pre-Trained Self-Supervised Vision Models](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.16674) (George Cazenavette et al., NeurIPS 2025) [:globe_with_meridians:](https:\u002F\u002Flinear-gradient-matching.github.io\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002FGeorgeCazenavette\u002Flinear-gradient-matching) [:book:](.\u002Fcitations\u002Fcazenavette2025dataset.txt)\n\n\u003Ca name=\"benchmark\" \u002F>\n\n### Benchmark\n\n+ [DC-BENCH: Dataset Condensation Benchmark](https:\u002F\u002Farxiv.org\u002Fabs\u002F2207.09639) (Justin Cui et al., NeurIPS 2022) [:globe_with_meridians:](https:\u002F\u002Fdc-bench.github.io\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fjustincui03\u002Fdc_benchmark) [:book:](.\u002Fcitations\u002Fcui2022dc.txt)\n+ [A Comprehensive Study on Dataset Distillation: Performance, Privacy, Robustness and Fairness](https:\u002F\u002Farxiv.org\u002Fabs\u002F2305.03355) (Zongxiong Chen & Jiahui Geng et al., 2023) [:book:](.\u002Fcitations\u002Fchen2023study.txt)\n+ [BEARD: Benchmarking the Adversarial Robustness for Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2411.09265) (Zheng Zhou et al., 2024) [:globe_with_meridians:](https:\u002F\u002Fbeard-leaderboard.github.io\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fzhouzhengqd\u002FBEARD\u002F) [:book:](.\u002Fcitations\u002Fzhou2024beard.txt)\n+ [DD-RobustBench: An Adversarial Robustness Benchmark for Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2403.13322) (Yifan Wu et al., TIP 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002FFredWU-HUST\u002FDD-RobustBench) [:book:](.\u002Fcitations\u002Fwu2025robust.txt)\n+ [DD-Ranking: Rethinking the Evaluation of Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.13300) (Zekai Li & Xinhao Zhong et al., 2025) [:globe_with_meridians:](https:\u002F\u002Fnus-hpc-ai-lab.github.io\u002FDD-Ranking\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002FNUS-HPC-AI-Lab\u002FDD-Ranking) [:book:](.\u002Fcitations\u002Fli2025ranking.txt)\n+ [Rectified Decoupled Dataset Distillation: A Closer Look for Fair and Comprehensive Evaluation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2509.19743) (Xinhao Zhong et al., ICLR 2026) [:book:](.\u002Fcitations\u002Fzhong2026rd3.txt)\n\n\u003Ca name=\"survey\" \u002F>\n\n### Survey\n\n+ [Data Distillation: A Survey](https:\u002F\u002Farxiv.org\u002Fabs\u002F2301.04272) (Noveen Sachdeva et al., TMLR 2023) [:book:](.\u002Fcitations\u002Fsachdeva2023survey.txt)\n+ [A Survey on Dataset Distillation: Approaches, Applications and Future Directions](https:\u002F\u002Farxiv.org\u002Fabs\u002F2305.01975) (Jiahui Geng & Zongxiong Chen et al., IJCAI 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002FGuang000\u002FAwesome-Dataset-Distillation) [:book:](.\u002Fcitations\u002Fgeng2023survey.txt)\n+ [A Comprehensive Survey to Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2301.05603) (Shiye Lei et al., TPAMI 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002FGuang000\u002FAwesome-Dataset-Distillation) [:book:](.\u002Fcitations\u002Flei2023survey.txt)\n+ [Dataset Distillation: A Comprehensive Review](https:\u002F\u002Farxiv.org\u002Fabs\u002F2301.07014) (Ruonan Yu & Songhua Liu et al., TPAMI 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002FGuang000\u002FAwesome-Dataset-Distillation) [:book:](.\u002Fcitations\u002Fyu2023review.txt)\n+ [The Evolution of Dataset Distillation: Toward Scalable and Generalizable Solutions](https:\u002F\u002Farxiv.org\u002Fabs\u002F2502.05673) (Ping Liu et al., 2025) [:book:](.\u002Fcitations\u002Fliu2025survey.txt)\n\n\u003Ca name=\"thesis\" \u002F>\n\n### Ph.D. Thesis\n+ [Data-efficient Neural Network Training with Dataset Condensation](https:\u002F\u002Fera.ed.ac.uk\u002Fhandle\u002F1842\u002F39756) (Bo Zhao, The University of Edinburgh 2023) [:book:](.\u002Fcitations\u002Fzhao2023thesis.txt)\n\n\u003Ca name=\"workshop\" \u002F>\n\n### Workshop\n+ 1st CVPR Workshop on Dataset Distillation (Saeed Vahidian et al., CVPR 2024) [:globe_with_meridians:](https:\u002F\u002Fsites.google.com\u002Fview\u002Fdd-cvpr2024\u002Fhome)\n\n\u003Ca name=\"challenge\" \u002F>\n\n### Challenge\n+ The First Dataset Distillation Challenge (Kai Wang & Ahmad Sajedi et al., ECCV 2024) [:globe_with_meridians:](https:\u002F\u002Fwww.dd-challenge.com\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002FDataDistillation\u002FECCV2024-Dataset-Distillation-Challenge)\n\n## Applications\n\n\u003Ca name=\"continual\" \u002F>\n\n### Continual Learning\n+ [Reducing Catastrophic Forgetting with Learning on Synthetic Data](https:\u002F\u002Farxiv.org\u002Fabs\u002F2004.14046) (Wojciech Masarczyk et al., CVPR 2020 Workshop) [:book:](.\u002Fcitations\u002Fmasarczyk2020reducing.txt)\n+ [Condensed Composite Memory Continual Learning](https:\u002F\u002Farxiv.org\u002Fabs\u002F2102.09890) (Felix Wiewel et al., IJCNN 2021) [:octocat:](https:\u002F\u002Fgithub.com\u002FFelixWiewel\u002FCCMCL) [:book:](.\u002Fcitations\u002Fwiewel2021soft.txt)\n+ [Distilled Replay: Overcoming Forgetting through Synthetic Samples](https:\u002F\u002Farxiv.org\u002Fabs\u002F2103.15851) (Andrea Rosasco et al., IJCAI 2021 Workshop) [:octocat:](https:\u002F\u002Fgithub.com\u002Fandrearosasco\u002FDistilledReplay) [:book:](.\u002Fcitations\u002Frosasco2021distilled.txt)\n+ [Sample Condensation in Online Continual Learning](https:\u002F\u002Farxiv.org\u002Fabs\u002F2206.11849) (Mattia Sangermano et al., IJCNN 2022) [:octocat:](https:\u002F\u002Fgithub.com\u002FMattiaSangermano\u002FOLCGM) [:book:](.\u002Fcitations\u002Fsangermano2022sample.txt)\n+ [An Efficient Dataset Condensation Plugin and Its Application to Continual Learning](https:\u002F\u002Fopenreview.net\u002Fforum?id=Murj6wcjRw) (Enneng Yang et al., NeurIPS 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002FEnnengYang\u002FAn-Efficient-Dataset-Condensation-Plugin) [:book:](.\u002Fcitations\u002Fyang2023efficient.txt)\n+ [Summarizing Stream Data for Memory-Restricted Online Continual Learning](https:\u002F\u002Farxiv.org\u002Fabs\u002F2305.16645) (Jianyang Gu et al., AAAI 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fvimar-gu\u002FSSD) [:book:](.\u002Fcitations\u002Fgu2024ssd.txt)\n+ [CD2: Constrained Dataset Distillation for Few-Shot Class-Incremental Learning](https:\u002F\u002Farxiv.org\u002Fabs\u002F2601.08519) (Kexin Bao et al., IJCAI 2025) [:book:](.\u002Fcitations\u002Fbao2025cd2.txt)\n+ [Asymmetric Synthetic Data Update for Domain Incremental Dataset Distillation](https:\u002F\u002Fopenreview.net\u002Fforum?id=XcsaCHaoJh) (Minyoung Oh et al., ICLR 2026) [:octocat:](https:\u002F\u002Fgithub.com\u002Fmyoh97\u002FDIDD-ASU) [:book:](.\u002Fcitations\u002Foh2026asu.txt)\n\n\u003Ca name=\"privacy\" \u002F>\n\n### Privacy\n+ [Privacy for Free: How does Dataset Condensation Help Privacy?](https:\u002F\u002Farxiv.org\u002Fabs\u002F2206.00240) (Tian Dong et al., ICML 2022) [:book:](.\u002Fcitations\u002Fdong2022privacy.txt)\n+ [Private Set Generation with Discriminative Information](https:\u002F\u002Farxiv.org\u002Fabs\u002F2211.04446) (Dingfan Chen et al., NeurIPS 2022) [:octocat:](https:\u002F\u002Fgithub.com\u002FDingfanChen\u002FPrivate-Set) [:book:](.\u002Fcitations\u002Fchen2022privacy.txt)\n+ [No Free Lunch in \"Privacy for Free: How does Dataset Condensation Help Privacy\"](https:\u002F\u002Farxiv.org\u002Fabs\u002F2209.14987) (Nicholas Carlini et al., 2022) [:book:](.\u002Fcitations\u002Fcarlini2022no.txt)\n+ [Backdoor Attacks Against Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2301.01197) (Yugeng Liu et al., NDSS 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Fliuyugeng\u002Fbaadd) [:book:](.\u002Fcitations\u002Fliu2023backdoor.txt)\n+ [Differentially Private Kernel Inducing Points (DP-KIP) for Privacy-preserving Data Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2301.13389) (Margarita Vinaroz et al., 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Fdpclip\u002Fdpclip) [:book:](.\u002Fcitations\u002Fvinaroz2023dpkip.txt)\n+ [Understanding Reconstruction Attacks with the Neural Tangent Kernel and Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2302.01428) (Noel Loo et al., ICLR 2024) [:book:](.\u002Fcitations\u002Floo2024attack.txt)\n+ [Rethinking Backdoor Attacks on Dataset Distillation: A Kernel Method Perspective](https:\u002F\u002Farxiv.org\u002Fabs\u002F2311.16646) (Ming-Yu Chung et al., ICLR 2024) [:book:](.\u002Fcitations\u002Fchung2024backdoor.txt)\n+ [Differentially Private Dataset Condensation](https:\u002F\u002Fwww.ndss-symposium.org\u002Fndss-paper\u002Fauto-draft-542\u002F) (Zheng et al., NDSS 2024 Workshop) [:book:](.\u002Fcitations\u002Fzheng2024differentially.txt)\n+ [Adaptive Backdoor Attacks Against Dataset Distillation for Federated Learning](https:\u002F\u002Fieeexplore.ieee.org\u002Fabstract\u002Fdocument\u002F10622462?casa_token=tHyZ-Pz7DpUAAAAA:vmCYI4cUcKzMluUsASHhIhr0CvBkjzBR-0N7REVj7aFN5hT5TinQTpSEsE0Bo3Fl8auh52Fipm_v) (Ze Chai et al., ICC 2024) [:book:](.\u002Fcitations\u002Fchai2024backdoor.txt)\n+ [Improving Noise Efficiency in Privacy-preserving Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2508.01749) (Runkai Zheng et al., ICCV 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002Fhumansensinglab\u002FDosser) [:book:](.\u002Fcitations\u002Fzheng2025dosser.txt)\n+ [SNEAKDOOR: Stealthy Backdoor Attacks against Distribution Matching-based Dataset Condensation](https:\u002F\u002Fopenreview.net\u002Fforum?id=YWE9na9Jai) (He Yang & Dongyi Lv et al., NeurIPS 2025) [:book:](.\u002Fcitations\u002Fyang2025sneakdoor.txt)\n+ [Poisoned Distillation: Injecting Backdoors into Distilled Datasets Without Raw Data Access](https:\u002F\u002Farxiv.org\u002Fabs\u002F2502.04229) (Ziyuan Yang et al., AAAI 2026) [:book:](.\u002Fcitations\u002Fyang2026pd.txt)\n+ [DP-GENG: Differentially Private Dataset Distillation Guided by DP-Generated Data](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.09876) (Shuo Shi et al., AAAI 2026) [:book:](.\u002Fcitations\u002Fshi2026dpgeng.txt)\n\n\u003Ca name=\"medical\" \u002F>\n\n### Medical\n+ [Soft-Label Anonymous Gastric X-ray Image Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2104.02857) (Guang Li et al., ICIP 2020) [:octocat:](https:\u002F\u002Fgithub.com\u002FGuang000\u002Fdataset-distillation) [:book:](.\u002Fcitations\u002Fli2020soft.txt) \n+ [Compressed Gastric Image Generation Based on Soft-Label Dataset Distillation for Medical Data Sharing](https:\u002F\u002Farxiv.org\u002Fabs\u002F2209.14635) (Guang Li et al., CMPB 2022) [:octocat:](https:\u002F\u002Fgithub.com\u002FGuang000\u002Fdataset-distillation) [:book:](.\u002Fcitations\u002Fli2022compressed.txt)\n+ [Dataset Distillation for Medical Dataset Sharing](https:\u002F\u002Fr2hcai.github.io\u002FAAAI-23\u002Fpages\u002Faccepted-papers.html) (Guang Li et al., AAAI 2023 Workshop) [:octocat:](https:\u002F\u002Fgithub.com\u002FGuang000\u002Fmtt-distillation) [:book:](.\u002Fcitations\u002Fli2023sharing.txt)\n+ [Communication-Efficient Federated Skin Lesion Classification with Generalizable Dataset Distillation](https:\u002F\u002Flink.springer.com\u002Fchapter\u002F10.1007\u002F978-3-031-47401-9_2) (Yuchen Tian & Jiacheng Wang et al., MICCAI 2023 Workshop) [:book:](.\u002Fcitations\u002Ftian2023gdd.txt)\n+ [Importance-Aware Adaptive Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2401.15863) (Guang Li et al., NN 2024) [:book:](.\u002Fcitations\u002Fli2024iadd.txt)\n+ [Image Distillation for Safe Data Sharing in Histopathology](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.13536) (Zhe Li et al., MICCAI 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FZheLi2020\u002FInfoDist) [:book:](.\u002Fcitations\u002Fli2024infodist.txt)\n+ [MedSynth: Leveraging Generative Model for Healthcare Data Sharing](https:\u002F\u002Flink.springer.com\u002Fchapter\u002F10.1007\u002F978-3-031-72390-2_61) (Renuga Kanagavelu et al., MICCAI 2024) [:book:](.\u002Fcitations\u002Fkanagavelu2024medsynth.txt)\n+ [Progressive Trajectory Matching for Medical Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2403.13469) (Zhen Yu et al., 2024) [:book:](.\u002Fcitations\u002Fyu2024progressive.txt)\n+ [Dataset Distillation in Medical Imaging: A Feasibility Study](https:\u002F\u002Farxiv.org\u002Fabs\u002F2407.14429) (Muyang Li et al., 2024) [:book:](.\u002Fcitations\u002Fli2024medical.txt)\n+ [Dataset Distillation for Histopathology Image Classification](https:\u002F\u002Farxiv.org\u002Fabs\u002F2408.09709) (Cong Cong et al., 2024) [:book:](.\u002Fcitations\u002Fcong2024dataset.txt)\n+ [Multi-modal Vision Pre-training for Medical Image Analysis](https:\u002F\u002Farxiv.org\u002Fabs\u002F2410.10604) (Shaohao Rui & Lingzhi Chen et al., CVPR 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002Fopenmedlab\u002FBrainMVP) [:book:](.\u002Fcitations\u002Frui2025brain.txt)\n+ [FedWSIDD: Federated Whole Slide Image Classification via Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2506.15365) (Haolong Jin et al., MICCAI 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002Ff1oNae\u002FFedWSIDD) [:book:](.\u002Fcitations\u002Fjin2025fedwsidd.txt)\n+ [High-Order Progressive Trajectory Matching for Medical Image Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2509.24177) (Le Dong et al., MICCAI 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002FBian-jh\u002FHoP-TM) [:book:](.\u002Fcitations\u002Fdong2025hop.txt)\n+ [Low-Level Dataset Distillation for Medical Image Enhancemen](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.13106) (Fengzhi Xu et al., 2025) [:book:](.\u002Fcitations\u002Fxu2025low.txt)\n\n\u003Ca name=\"fed\" \u002F>\n\n### Federated Learning\n+ [Federated Learning via Synthetic Data](https:\u002F\u002Farxiv.org\u002Fabs\u002F2008.04489) (Jack Goetz et al., 2020) [:book:](.\u002Fcitations\u002Fgoetz2020federated.txt)\n+ [Distilled One-Shot Federated Learning](https:\u002F\u002Farxiv.org\u002Fabs\u002F2009.07999) (Yanlin Zhou et al., 2020) [:book:](.\u002Fcitations\u002Fzhou2020distilled.txt)\n+ [DENSE: Data-Free One-Shot Federated Learning](https:\u002F\u002Farxiv.org\u002Fabs\u002F2112.12371) (Jie Zhang & Chen Chen et al., NeurIPS 2022) [:octocat:](https:\u002F\u002Fgithub.com\u002Fzj-jayzhang\u002FDENSE) [:book:](.\u002Fcitations\u002Fzhang2022dense.txt)\n+ [FedSynth: Gradient Compression via Synthetic Data in Federated Learning](https:\u002F\u002Farxiv.org\u002Fabs\u002F2204.01273) (Shengyuan Hu et al., 2022) [:book:](.\u002Fcitations\u002Fhu2022fedsynth.txt)\n+ [Meta Knowledge Condensation for Federated Learning](https:\u002F\u002Farxiv.org\u002Fabs\u002F2209.14851) (Ping Liu et al., ICLR 2023) [:book:](.\u002Fcitations\u002Fliu2023meta.txt)\n+ [DYNAFED: Tackling Client Data Heterogeneity with Global Dynamics](https:\u002F\u002Farxiv.org\u002Fabs\u002F2211.10878) (Renjie Pi et al., CVPR 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Fpipilurj\u002Fdynafed) [:book:](.\u002Fcitations\u002Fpi2023dynafed.txt)\n+ [FedDM: Iterative Distribution Matching for Communication-Efficient Federated Learning](https:\u002F\u002Farxiv.org\u002Fabs\u002F2207.09653) (Yuanhao Xiong & Ruochen Wang et al., CVPR 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Fanonymifish\u002Ffed-distribution-matching) [:book:](.\u002Fcitations\u002Fxiong2023feddm.txt)\n+ [Federated Learning via Decentralized Dataset Distillation in Resource-Constrained Edge Environments](https:\u002F\u002Farxiv.org\u002Fabs\u002F2208.11311) (Rui Song et al., IJCNN 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Frruisong\u002Ffedd3) [:book:](.\u002Fcitations\u002Fsong2023federated.txt)\n+ [FedLAP-DP: Federated Learning by Sharing Differentially Private Loss Approximations](https:\u002F\u002Farxiv.org\u002Fabs\u002F2302.01068) (Hui-Po Wang et al., 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Fa514514772\u002Ffedlap-dp) [:book:](.\u002Fcitations\u002Fwang2023fed.txt)\n+ [Federated Learning on Virtual Heterogeneous Data with Local-global Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2303.02278) (Chun-Yin Huang et al., TMLR 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fubc-tea\u002FFedLGD) [:book:](.\u002Fcitations\u002Fhuang2024federated.txt)\n+ [An Aggregation-Free Federated Learning for Tackling Data Heterogeneity](https:\u002F\u002Farxiv.org\u002Fabs\u002F2404.18962) (Yuan Wang et al., CVPR 2024) [:book:](.\u002Fcitations\u002Fwang2024fed.txt)\n+ [Overcoming Data and Model Heterogeneities in Decentralized Federated Learning via Synthetic Anchors](https:\u002F\u002Farxiv.org\u002Fabs\u002F2405.11525) (Chun-Yin Huang et al., ICML 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fubc-tea\u002FDESA) [:book:](.\u002Fcitations\u002Fhuang2024desa.txt)\n+ [DCFL: Non-IID Awareness Dataset Condensation Aided Federated Learning](https:\u002F\u002Fieeexplore.ieee.org\u002Fdocument\u002F10650791) (Xingwang Wang et al., IJCNN 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FJLUssh\u002FDCFL) [:book:](.\u002Fcitations\u002Fwang2024dcfl.txt)\n+ [Unlocking the Potential of Federated Learning: The Symphony of Dataset Distillation via Deep Generative Latents](https:\u002F\u002Farxiv.org\u002Fabs\u002F2312.01537) (Yuqi Jia & Saeed Vahidian et al., ECCV 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FFedDG23\u002FFedDG-main) [:book:](.\u002Fcitations\u002Fjia2024feddg.txt)\n+ [One-Shot Collaborative Data Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2408.02266) (William Holland et al., ECAI 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Frayneholland\u002FCollabDM) [:book:](.\u002Fcitations\u002Fholland2024one.txt)\n+ [FedVCK: Non-IID Robust and Communication-Efficient Federated Learning via Valuable Condensed Knowledge for Medical Image Analysis](https:\u002F\u002Farxiv.org\u002Fabs\u002F2412.18557) (Guochen Yan et al., AAAI 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002FYouth-49\u002FFedVCK_2024) [:book:](.\u002Fcitations\u002Fyan2025fedvck.txt)\n\n\n\u003Ca name=\"gnn\" \u002F>\n\n### Graph Neural Network\n+ [Graph Condensation for Graph Neural Networks](https:\u002F\u002Farxiv.org\u002Fabs\u002F2110.07580) (Wei Jin et al., ICLR 2022) [:octocat:](https:\u002F\u002Fgithub.com\u002Fchandlerbang\u002Fgcond) [:book:](.\u002Fcitations\u002Fjin2022graph.txt)\n+ [Condensing Graphs via One-Step Gradient Matching](https:\u002F\u002Farxiv.org\u002Fabs\u002F2206.07746) (Wei Jin et al., KDD 2022) [:octocat:](https:\u002F\u002Fgithub.com\u002Famazon-research\u002FDosCond) [:book:](.\u002Fcitations\u002Fjin2022condensing.txt)\n+ [Graph Condensation via Receptive Field Distribution Matching](https:\u002F\u002Farxiv.org\u002Fabs\u002F2206.13697) (Mengyang Liu et al., 2022) [:book:](.\u002Fcitations\u002Fliu2022graph.txt)\n+ [Kernel Ridge Regression-Based Graph Dataset Distillation](https:\u002F\u002Fdl.acm.org\u002Fdoi\u002F10.1145\u002F3580305.3599398) (Zhe Xu et al., KDD 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Fpricexu\u002FKIDD) [:book:](.\u002Fcitations\u002Fxu2023kidd.txt)\n+ [Structure-free Graph Condensation: From Large-scale Graphs to Condensed Graph-free Data](https:\u002F\u002Farxiv.org\u002Fabs\u002F2306.02664) (Xin Zheng et al., NeurIPS 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Famanda-zheng\u002Fsfgc) [:book:](.\u002Fcitations\u002Fzheng2023sfgc.txt)\n+ [Does Graph Distillation See Like Vision Dataset Counterpart?](https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.09192) (Beining Yang & Kai Wang et al., NeurIPS 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002FRingBDStack\u002FSGDD) [:book:](.\u002Fcitations\u002Fyang2023sgdd.txt)\n+ [CaT: Balanced Continual Graph Learning with Graph Condensation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2309.09455) (Yilun Liu et al., ICDM 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Fsuperallen13\u002FCaT-CGL) [:book:](.\u002Fcitations\u002Fliu2023cat.txt)\n+ [Mirage: Model-Agnostic Graph Distillation for Graph Classification](https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.09486) (Mridul Gupta & Sahil Manchanda et al., ICLR 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Ffrigategnn\u002FMirage) [:book:](.\u002Fcitations\u002Fgupta2024mirage.txt)\n+ [Graph Distillation with Eigenbasis Matching](https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.09202) (Yang Liu & Deyu Bo et al., ICML 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fliuyang-tian\u002FGDEM) [:book:](.\u002Fcitations\u002Fliu2024gdem.txt)\n+ [Navigating Complexity: Toward Lossless Graph Condensation via Expanding Window Matching](https:\u002F\u002Farxiv.org\u002Fabs\u002F2402.05011) (Yuchen Zhang & Tianle Zhang & Kai Wang et al., ICML 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fnus-hpc-ai-lab\u002Fgeom) [:book:](.\u002Fcitations\u002Fzhang2024geom.txt)\n+ [Graph Data Condensation via Self-expressive Graph Structure Reconstruction](https:\u002F\u002Farxiv.org\u002Fabs\u002F2403.07294) (Zhanyu Liu & Chaolv Zeng et al., KDD 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fzclzcl0223\u002FGCSR) [:book:](.\u002Fcitations\u002Fliu2024gcsr.txt)\n+ [Two Trades is not Baffled: Condensing Graph via Crafting Rational Gradient Matching](https:\u002F\u002Farxiv.org\u002Fabs\u002F2402.04924) (Tianle Zhang & Yuchen Zhang & Kai Wang et al., 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fnus-hpc-ai-lab\u002Fctrl) [:book:](.\u002Fcitations\u002Fzhang2024ctrl.txt)\n\n#### Survey\n+ [A Comprehensive Survey on Graph Reduction: Sparsification, Coarsening, and Condensation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2402.03358) (Mohammad Hashemi et al., IJCAI 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FEmory-Melody\u002Fawesome-graph-reduction) [:book:](.\u002Fcitations\u002Fhashemi2024awesome.txt)\n+ [A Survey on Graph Condensation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2402.02000) (Hongjia Xu et al., 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FFrostland12138\u002FAwesome-Graph-Condensation) [:book:](.\u002Fcitations\u002Fxu2024survey.txt)\n+ [Graph Condensation: A Survey](https:\u002F\u002Farxiv.org\u002Fabs\u002F2401.11720) (Xinyi Gao et al., TKDE 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002Fxygaog\u002Fgraph-condensation-papers) [:book:](.\u002Fcitations\u002Fgao2025graph.txt)\n\n#### Benchmark\n+ [GC-Bench: An Open and Unified Benchmark for Graph Condensation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2407.00615) (Qingyun Sun & Ziying Chen et al., NeurIPS 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FRingBDStack\u002FGC-Bench) [:book:](.\u002Fcitations\u002Fsun2024gcbench.txt)\n+ [GCondenser: Benchmarking Graph Condensation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2405.14246) (Yilun Liu et al., 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fsuperallen13\u002FGCondenser) [:book:](.\u002Fcitations\u002Fliu2024gcondenser.txt)\n+ [GC-Bench: A Benchmark Framework for Graph Condensation with New Insights](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.16715) (Shengbo Gong & Juntong Ni et al., 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FEmory-Melody\u002FGraphSlim) [:book:](.\u002Fcitations\u002Fgong2024graphslim.txt)\n\n#### No further updates will be made regarding graph distillation topics as sufficient papers and summary projects are already available on the subject\n\n\u003Ca name=\"nas\" \u002F>\n\n### Neural Architecture Search\n+ [Generative Teaching Networks: Accelerating Neural Architecture Search by Learning to Generate Synthetic Training Data](https:\u002F\u002Farxiv.org\u002Fabs\u002F1912.07768) (Felipe Petroski Such et al., ICML 2020) [:octocat:](https:\u002F\u002Fgithub.com\u002Fuber-research\u002FGTN) [:book:](.\u002Fcitations\u002Fsuch2020generative.txt)\n+ [Learning to Generate Synthetic Training Data using Gradient Matching and Implicit Differentiation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2203.08559) (Dmitry Medvedev et al., AIST 2021) [:octocat:](https:\u002F\u002Fgithub.com\u002Fdm-medvedev\u002Fefficientdistillation) [:book:](.\u002Fcitations\u002Fmedvedev2021tabular.txt)\n+ [Calibrated Dataset Condensation for Faster Hyperparameter Search](https:\u002F\u002Farxiv.org\u002Fabs\u002F2405.17535) (Mucong Ding et al., 2024) [:book:](.\u002Fcitations\u002Fding2024hcdc.txt)\n\n\u003Ca name=\"fashion\" \u002F>\n\n### Fashion, Art, and Design\n+ [Wearable ImageNet: Synthesizing Tileable Textures via Dataset Distillation](https:\u002F\u002Fopenaccess.thecvf.com\u002Fcontent\u002FCVPR2022W\u002FCVFAD\u002Fhtml\u002FCazenavette_Wearable_ImageNet_Synthesizing_Tileable_Textures_via_Dataset_Distillation_CVPRW_2022_paper.html) (George Cazenavette et al., CVPR 2022 Workshop) [:globe_with_meridians:](https:\u002F\u002Fgeorgecazenavette.github.io\u002Fmtt-distillation\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fgeorgecazenavette\u002Fmtt-distillation) [:book:](.\u002Fcitations\u002Fcazenavette2022textures.txt)\n+ [Learning from Designers: Fashion Compatibility Analysis Via Dataset Distillation](https:\u002F\u002Fieeexplore.ieee.org\u002Fdocument\u002F9897234) (Yulan Chen et al., ICIP 2022) [:book:](.\u002Fcitations\u002Fchen2022fashion.txt)\n+ [Galaxy Dataset Distillation with Self-Adaptive Trajectory Matching](https:\u002F\u002Farxiv.org\u002Fabs\u002F2311.17967) (Haowen Guan et al., NeurIPS 2023 Workshop) [:octocat:](https:\u002F\u002Fgithub.com\u002FHaowenGuan\u002FGalaxy-Dataset-Distillation) [:book:](.\u002Fcitations\u002Fguan2023galaxy.txt)\n\n\u003Ca name=\"rec\" \u002F>\n\n### Recommender Systems\n+ [Infinite Recommendation Networks: A Data-Centric Approach](https:\u002F\u002Farxiv.org\u002Fabs\u002F2206.02626) (Noveen Sachdeva et al., NeurIPS 2022) [:octocat:](https:\u002F\u002Fgithub.com\u002Fnoveens\u002Fdistill_cf) [:book:](.\u002Fcitations\u002Fsachdeva2022data.txt)\n+ [Gradient Matching for Categorical Data Distillation in CTR Prediction](https:\u002F\u002Fdl.acm.org\u002Fdoi\u002F10.1145\u002F3604915.3608769) (Chen Wang et al., RecSys 2023) [:book:](.\u002Fcitations\u002Fwang2023cgm.txt)\n+ [TD3: Tucker Decomposition Based Dataset Distillation Method for Sequential Recommendation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2502.02854) (Jiaqing Zhang et al., WWW 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002FUSTC-StarTeam\u002FTD3) [:book:](.\u002Fcitations\u002Fzhang2025td3.txt)\n+ [DIET: Learning to Distill Dataset Continually for Recommender Systems](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.24958) (Jiaqing Zhang et al., 2026) [:book:](.\u002Fcitations\u002Fzhang2026diet.txt)\n\n\u003Ca name=\"blackbox\" \u002F>\n\n### Blackbox Optimization\n+ [Bidirectional Learning for Offline Infinite-width Model-based Optimization](https:\u002F\u002Farxiv.org\u002Fabs\u002F2209.07507) (Can Chen et al., NeurIPS 2022) [:octocat:](https:\u002F\u002Fgithub.com\u002Fggchen1997\u002Fbdi) [:book:](.\u002Fcitations\u002Fchen2022bidirectional.txt) \n+ [Bidirectional Learning for Offline Model-based Biological Sequence Design](https:\u002F\u002Farxiv.org\u002Fabs\u002F2301.02931) (Can Chen et al., ICML 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002FGGchen1997\u002FBIB-ICML2023-Submission) [:book:](.\u002Fcitations\u002Fchen2023bidirectional.txt)\n\n\u003Ca name=\"robustness\" \u002F>\n\n### Robustness\n+ [Can We Achieve Robustness from Data Alone?](https:\u002F\u002Farxiv.org\u002Fabs\u002F2207.11727) (Nikolaos Tsilivis et al., ICML 2022 Workshop) [:book:](.\u002Fcitations\u002Ftsilivis2022robust.txt)\n+ [Towards Robust Dataset Learning](https:\u002F\u002Farxiv.org\u002Fabs\u002F2211.10752) (Yihan Wu et al., 2022) [:book:](.\u002Fcitations\u002Fwu2022towards.txt)\n+ [Rethinking Data Distillation: Do Not Overlook Calibration](https:\u002F\u002Farxiv.org\u002Fabs\u002F2307.12463) (Dongyao Zhu et al., ICCV 2023) [:book:](.\u002Fcitations\u002Fzhu2023calibration.txt)\n+ [Towards Trustworthy Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2307.09165) (Shijie Ma et al., PR 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fmashijie1028\u002FTrustDD\u002F)  [:book:](.\u002Fcitations\u002Fma2024trustworthy.txt)\n+ [Towards Adversarially Robust Dataset Distillation by Curvature Regularization](https:\u002F\u002Farxiv.org\u002Fabs\u002F2403.10045) (Eric Xue et al., AAAI 2025) [:globe_with_meridians:](https:\u002F\u002Fyumozi.github.io\u002FGUARD\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fyumozi\u002FGUARD) [:book:](.\u002Fcitations\u002Fxue2025robust.txt)\n+ [Group Distributionally Robust Dataset Distillation with Risk Minimization](https:\u002F\u002Farxiv.org\u002Fabs\u002F2402.04676) (Saeed Vahidian & Mingyu Wang & Jianyang Gu et al., ICLR 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002FMming11\u002FRobustDatasetDistillation) [:book:](.\u002Fcitations\u002Fvahidian2025group.txt)\n+ [ROME is Forged in Adversity: Robust Distilled Datasets via Information Bottleneck](https:\u002F\u002Fopenreview.net\u002Fforum?id=agtwOsnLUB) (Zheng Zhou et al., ICML 2025) [:globe_with_meridians:](https:\u002F\u002Fzhouzhengqd.github.io\u002Frome.page\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fzhouzhengqd\u002FROME) [:book:](.\u002Fcitations\u002Fzhou2025rome.txt)\n\n\u003Ca name=\"fairness\" \u002F>\n\n### Fairness\n+ [Fair Graph Distillation](https:\u002F\u002Fopenreview.net\u002Fforum?id=xW0ayZxPWs) (Qizhang Feng et al., NeurIPS 2023) [:book:](.\u002Fcitations\u002Ffeng2023fair.txt)\n+ [FairDD: Fair Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2411.19623) (Qihang Zhou et al., NeurIPS 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002Fzqhang\u002FFairDD) [:book:](.\u002Fcitations\u002Fzhou2025fair.txt)\n\n\u003Ca name=\"text\" \u002F>\n\n### Text\n+ [Data Distillation for Text Classification](https:\u002F\u002Farxiv.org\u002Fabs\u002F2104.08448) (Yongqi Li et al., 2021) [:book:](.\u002Fcitations\u002Fli2021text.txt)\n+ [Dataset Distillation with Attention Labels for Fine-tuning BERT](https:\u002F\u002Faclanthology.org\u002F2023.acl-short.12\u002F) (Aru Maekawa et al., ACL 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Farumaekawa\u002Fdataset-distillation-with-attention-labels) [:book:](.\u002Fcitations\u002Fmaekawa2023text.txt)\n+ [DiLM: Distilling Dataset into Language Model for Text-level Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2404.00264) (Aru Maekawa et al., NAACL 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Farumaekawa\u002FDiLM) [:book:](.\u002Fcitations\u002Fmaekawa2024dilm.txt)\n+ [Textual Dataset Distillation via Language Model Embedding](https:\u002F\u002Faclanthology.org\u002F2024.findings-emnlp.733\u002F) (Yefan Tao et al., EMNLP 2024) [:book:](.\u002Fcitations\u002Ftao2024textual.txt)\n+ [UniDetox: Universal Detoxification of Large Language Models via Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2504.20500) (Huimin Lu et al., ICLR 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002FEminLU\u002FUniDetox) [:book:](.\u002Fcitations\u002Flu2025llm.txt)\n+ [Knowledge Hierarchy Guided Biological-Medical Dataset Distillation for Domain LLM Training](https:\u002F\u002Farxiv.org\u002Fabs\u002F2501.15108) (Xunxin Cai & Chengrui Wang & Qingqing Long et al., DASFAA 2025) [:book:](.\u002Fcitations\u002Fcai2025llm.txt)\n+ [Synthetic Text Generation for Training Large Language Models via Gradient Matching](https:\u002F\u002Farxiv.org\u002Fabs\u002F2502.17607) (Dang Nguyen & Zeman Li et al., ICML 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002FBigML-CS-UCLA\u002FGRADMM) [:book:](.\u002Fcitations\u002Fnguyen2025llm.txt)\n+ [CondenseLM: LLMs-driven Text Dataset Condensation via Reward Matching](https:\u002F\u002Faclanthology.org\u002F2025.emnlp-main.65\u002F) (Cheng Shen et al., EMNLP 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002Fcs6331\u002FCondenseLM\u002F) [:book:](.\u002Fcitations\u002Fshen2025llm.txt)\n\n\u003Ca name=\"video\" \u002F>\n\n### Video\n+ [Dancing with Still Images: Video Distillation via Static-Dynamic Disentanglement](https:\u002F\u002Farxiv.org\u002Fabs\u002F2312.00362) (Ziyu Wang & Yue Xu et al., CVPR 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fyuz1wan\u002Fvideo_distillation) [:book:](.\u002Fcitations\u002Fwang2023dancing.txt)\n+ [Video Set Distillation: Information Diversification and Temporal Densifica](https:\u002F\u002Farxiv.org\u002Fabs\u002F2412.00111) (Yinjie Zhao et al., 2024) [:book:](.\u002Fcitations\u002Fzhao2024video.txt)\n+ [A Large-Scale Study on Video Action Dataset Condensation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2412.21197) (Yang Chen et al., 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FMCG-NJU\u002FVideo-DC) [:book:](.\u002Fcitations\u002Fchen2024video.txt)\n+ [Condensing Action Segmentation Datasets via Generative Network Inversion](https:\u002F\u002Farxiv.org\u002Fabs\u002F2503.14112) (Guodong Ding et al., CVPR 2025) [:book:](.\u002Fcitations\u002Fding2025video.txt)\n+ [Latent Video Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2504.17132) (Ning Li et al., CVPR 2025 Workshop) [:octocat:](https:\u002F\u002Fgithub.com\u002Fliningresearch\u002FLatent_Video_Dataset_Distillation) [:book:](.\u002Fcitations\u002Fli2025latent.txt)\n+ [Distill Video Datasets into Images](https:\u002F\u002Farxiv.org\u002Fabs\u002F2512.14621) (Zhenghao Zhao et al., 2025) [:book:](.\u002Fcitations\u002Fzhao2025video.txt)\n+ [PRISM: Video Dataset Condensation with Progressive Refinement and Insertion for Sparse Motion](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.22564) (Jaehyun Choi et al., CVPR 2026) [:book:](.\u002Fcitations\u002Fchoi2026prism.txt)\n\n\u003Ca name=\"tabular\" \u002F>\n\n### Tabular\n+ [New Properties of the Data Distillation Method When Working With Tabular Data](https:\u002F\u002Farxiv.org\u002Fabs\u002F2010.09839) (Dmitry Medvedev et al., AIST 2020) [:octocat:](https:\u002F\u002Fgithub.com\u002Fdm-medvedev\u002Fdataset-distillation) [:book:](.\u002Fcitations\u002Fmedvedev2020tabular.txt)\n\n\u003Ca name=\"retrieval\" \u002F>\n\n### Retrieval\n+ [Towards Efficient Deep Hashing Retrieval: Condensing Your Data via Feature-Embedding Matching](https:\u002F\u002Farxiv.org\u002Fabs\u002F2305.18076) (Tao Feng & Jie Zhang et al., 2023) [:book:](.\u002Fcitations\u002Ffeng2023hash.txt)\n\n\u003Ca name=\"domain\" \u002F>\n\n### Domain Adaptation\n+ [Multi-Source Domain Adaptation Meets Dataset Distillation through Dataset Dictionary Learning](https:\u002F\u002Farxiv.org\u002Fabs\u002F2309.07666) (Eduardo Montesuma et al., ICASSP 2024) [:book:](.\u002Fcitations\u002Fmontesuma2024multi.txt)\n\n\u003Ca name=\"super\" \u002F>\n\n### Super Resolution\n+ [GSDD: Generative Space Dataset Distillation for Image Super-resolution](https:\u002F\u002Fojs.aaai.org\u002Findex.php\u002FAAAI\u002Farticle\u002Fview\u002F28534) (Haiyu Zhang et al., AAAI 2024) [:book:](.\u002Fcitations\u002Fzhang2024super.txt)\n\n\u003Ca name=\"time\" \u002F>\n\n### Time Series\n+ [Dataset Condensation for Time Series Classification via Dual Domain Matching](https:\u002F\u002Farxiv.org\u002Fabs\u002F2403.07245) (Zhanyu Liu et al., KDD 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fzhyliu00\u002FTimeSeriesCond) [:book:](.\u002Fcitations\u002Fliu2024time.txt)\n+ [CondTSF: One-line Plugin of Dataset Condensation for Time Series Forecasting](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.02131) (Jianrong Ding & Zhanyu Liu et al., NeurIPS 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FRafaDD\u002FCondTSF) [:book:](.\u002Fcitations\u002Fding2024time.txt)\n+ [Less is More: Efficient Time Series Dataset Condensation via Two-fold Modal Matching](https:\u002F\u002Farxiv.org\u002Fabs\u002F2410.20905) (Hao Miao et al., VLDB 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002Fuestc-liuzq\u002FSTdistillation) [:book:](.\u002Fcitations\u002Fmiao2025timedc.txt)\n+ [DDTime: Dataset Distillation with Spectral Alignment and Information Bottleneck for Time-Series Forecasting](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.16715) (Yuqi Li & Kuiye Ding et al., 2025) [:book:](.\u002Fcitations\u002Fli2025time.txt)\n+ [Harmonic Dataset Distillation for Time Series Forecasting](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.03760) (Seungha Hong et al., AAAI 2026) [:book:](.\u002Fcitations\u002Fhong2026hdt.txt)\n+ [Distilling Time Series Foundation Models for Efficient Forecasting](https:\u002F\u002Farxiv.org\u002Fabs\u002F2601.12785) (Yuqi Li & Kuiye Ding et al., ICASSP 2026) [:octocat:](https:\u002F\u002Fgithub.com\u002Fitsnotacie\u002FDistilTS-ICASSP2026) [:book:](.\u002Fcitations\u002Fli2026distilts.txt)\n+ [Effective Dataset Distillation for Spatio-Temporal Forecasting with Bi-dimensional Compression](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.10410) (Taehyung Kwon & Yeonje Choi et al., ICDE 2026) [:octocat:](https:\u002F\u002Fgithub.com\u002Fkbrother\u002FSTemDist) [:book:](.\u002Fcitations\u002Fkwon2026effective.txt)\n+ [ShapeCond: Fast Shapelet-Guided Dataset Condensation for Time Series Classification](https:\u002F\u002Farxiv.org\u002Fabs\u002F2602.09008) (Sijia Peng et al., 2026) [:octocat:](https:\u002F\u002Fgithub.com\u002Flunaaa95\u002FShapeCond) [:book:](.\u002Fcitations\u002Fpeng2026shapecond.txt)\n\n\u003Ca name=\"speech\" \u002F>\n\n### Speech\n+ [Dataset-Distillation Generative Model for Speech Emotion Recognition](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.02963) (Fabian Ritter-Gutierrez et al., Interspeech 2024) [:book:](.\u002Fcitations\u002Ffabian2024speech.txt)\n\n\u003Ca name=\"unlearning\" \u002F>\n\n### Machine Unlearning\n+ [Distilled Datamodel with Reverse Gradient Matching](https:\u002F\u002Farxiv.org\u002Fabs\u002F2404.14006) (Jingwen Ye et al., CVPR 2024) [:book:](.\u002Fcitations\u002Fye2024datamodel.txt)\n+ [Dataset Condensation Driven Machine Unlearning](https:\u002F\u002Farxiv.org\u002Fabs\u002F2402.00195) (Junaid Iqbal Khan, 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Falgebraicdianuj\u002FDC_U) [:book:](.\u002Fcitations\u002Fkhan2024unlearning.txt)\n\n\u003Ca name=\"rl\" \u002F>\n\n### Reinforcement Learning\n+ [Behaviour Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.15042) (Andrei Lupu et al., ICLR 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fflairox\u002Fbehaviour-distillation) [:book:](.\u002Fcitations\u002Flupu2024bd.txt)\n+ [Dataset Distillation for Offline Reinforcement Learning](https:\u002F\u002Farxiv.org\u002Fabs\u002F2407.20299) (Jonathan Light & Yuanzhe Liu et al., ICML 2024 Workshop) [:globe_with_meridians:](https:\u002F\u002Fdatasetdistillation4rl.github.io\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fggflow123\u002FDDRL) [:book:](.\u002Fcitations\u002Flight2024rl.txt)\n+ [Offline Behavior Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2410.22728) (Shiye Lei et al., NeurIPS 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FLeavesLei\u002FOBD) [:book:](.\u002Fcitations\u002Flei2024obl.txt)\n+ [Distilling Reinforcement Learning into Single-Batch Datasets](https:\u002F\u002Farxiv.org\u002Fabs\u002F2508.09283) (Connor Wilhelm et al., ECAI 2025) [:book:](.\u002Fcitations\u002Fwilhelm2025rl.txt)\n+ [Algorithmic Guarantees for Distilling Supervised and Offline RL Datasets](https:\u002F\u002Farxiv.org\u002Fabs\u002F2512.00536) (Aaryan Gupta et al., ICLR 2026) [:book:](.\u002Fcitations\u002Fgupta2026rl.txt)\n\n\u003Ca name=\"long\" \u002F>\n\n### Long-Tail\n+ [Distilling Long-tailed Datasets](https:\u002F\u002Farxiv.org\u002Fabs\u002F2408.14506) (Zhenghao Zhao & Haoxuan Wang et al., CVPR 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002Fichbill\u002FLTDD) [:book:](.\u002Fcitations\u002Fzhao2025long.txt)\n+ [Rectifying Soft-Label Entangled Bias in Long-Tailed Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.17914) (Chenyang Jiang et al., NeurIPS 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002Fj-cyoung\u002FADSA_DD) [:book:](.\u002Fcitations\u002Fjiang2025long.txt)\n+ [Rethinking Long-tailed Dataset Distillation: A Uni-Level Framework with Unbiased Recovery and Relabeling](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.18858) (Xiao Cui et al., AAAI 2026) [:book:](.\u002Fcitations\u002Fcui2026long.txt)\n\n\u003Ca name=\"noisy\" \u002F>\n\n### Learning with Noisy Labels\n+ [Dataset Distillers Are Good Label Denoisers In the Wild](https:\u002F\u002Farxiv.org\u002Fabs\u002F2411.11924) (Lechao Cheng et al., 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FKciiiman\u002FDD_LNL) [:book:](.\u002Fcitations\u002Fcheng2024noisy.txt)\n+ [Robust Dataset Condensation using Supervised Contrastive Learning](https:\u002F\u002Fopenaccess.thecvf.com\u002Fcontent\u002FICCV2025\u002Fhtml\u002FKim_Robust_Dataset_Condensation_using_Supervised_Contrastive_Learning_ICCV_2025_paper.html) (Nicole Hee-Yeon Kim et al., ICCV 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002FDISL-Lab\u002FRDC-ICCV2025) [:book:](.\u002Fcitations\u002Fkim2025rdc.txt)\n\n\u003Ca name=\"detection\" \u002F>\n\n### Object Detection\n+ [Fetch and Forge: Efficient Dataset Condensation for Object Detection](https:\u002F\u002Fopenreview.net\u002Fforum?id=m8MElyzuwp) (Ding Qi et al., NeurIPS 2024) [:book:](.\u002Fcitations\u002Fqi2024dcod.txt)\n+ [OD3: Optimization-free Dataset Distillation for Object Detection](https:\u002F\u002Farxiv.org\u002Fabs\u002F2506.01942) (Salwa K. Al Khatib & Ahmed ElHagry & Shitong Shao et al., ICLR 2026) [:octocat:](https:\u002F\u002Fgithub.com\u002FVILA-Lab\u002FOD3) [:book:](.\u002Fcitations\u002Fkhatib2026od3.txt)\n\n\u003Ca name=\"point\" \u002F>\n\n### Point Cloud\n+ [Point Cloud Dataset Distillation](https:\u002F\u002Fopenreview.net\u002Fforum?id=Us8v5tDOFd) (Deyu Bo et al., ICML 2025) [:book:](.\u002Fcitations\u002Fbo2025point.txt)\n+ [Dataset Distillation of 3D Point Clouds via Distribution Matching](https:\u002F\u002Farxiv.org\u002Fabs\u002F2503.22154) (Jae-Young Yim & Dongwook Kim et al., NeurIPS 2025) [:book:](.\u002Fcitations\u002Fyim2025point.txt)\n+ [Parameterization-Based Dataset Distillation of 3D Point Clouds through Learnable Shape Morphing](https:\u002F\u002Fopenreview.net\u002Fforum?id=Qe7dKZOtWM) (Dongwook Kim & Jae-Young Yim et al., ICLR 2026) [:book:](.\u002Fcitations\u002Fkim2026pointmorph.txt)\n\n\u003Ca name=\"uni\" \u002F>\n\n### Universal Distillation\n\n+ [Towards Universal Dataset Distillation via Task-Driven Diffusion](https:\u002F\u002Fopenaccess.thecvf.com\u002Fcontent\u002FCVPR2025\u002Fhtml\u002FQi_Towards_Universal_Dataset_Distillation_via_Task-Driven_Diffusion_CVPR_2025_paper.html) (Ding Qi et al., CVPR 2025) [:book:](.\u002Fcitations\u002Fqi2025unidd.txt)\n\n\u003Ca name=\"snn\" \u002F>\n\n### Spiking Neural Network\n\n+ [Learning from Dense Events: Towards Fast Spiking Neural Networks Training via Event Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.12095) (Shuhan Ye et al., 2025) [:book:](.\u002Fcitations\u002Fye2025snn.txt)\n\n\u003Ca name=\"eeg\" \u002F>\n\n### EEG\n\n+ [EEG-DLite: Dataset Distillation for Efficient Large EEG Model Training](https:\u002F\u002Farxiv.org\u002Fabs\u002F2512.12210) (Yuting Tang et al., AAAI 2026) [:octocat:](https:\u002F\u002Fgithub.com\u002Ft170815518\u002FEEG-DLite) [:book:](.\u002Fcitations\u002Ftang2026eeg.txt)\n\n\u003Ca name=\"finance\" \u002F>\n\n### Finance\n\n+ [Secure and Explainable Fraud Detection in Finance via Hierarchical Multi-source Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2512.21866) (Yiming Qian et al., ICAIFW 2025) [:book:](.\u002Fcitations\u002Fqian2025finance.txt)\n\n\u003Ca name=\"music\" \u002F>\n\n### Music\n\n+ [ConceptCaps: a Distilled Concept Dataset for Interpretability in Music Models](https:\u002F\u002Farxiv.org\u002Fabs\u002F2601.14157) (Bruno Sienkiewicz et al., 2026) [:book:](.\u002Fcitations\u002Fsienkiewicz2026music.txt)\n\n\u003Ca name=\"rs\" \u002F>\n\n### Remote Sensing\n\n+ [Towards Realistic Remote Sensing Dataset Distillation with Discriminative Prototype-guided Diffusion](https:\u002F\u002Farxiv.org\u002Fabs\u002F2601.15829) (Yonghao Xu et al., 2026) [:book:](.\u002Fcitations\u002Fxu2026rs.txt)\n\n\u003Ca name=\"dr\" \u002F>\n\n### Deraining\n\n+ [UniRain: Unified Image Deraining with RAG-based Dataset Distillation and Multi-objective Reweighted Optimization](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.03967) (Qianfeng Yang et al., CVPR 2026) [:octocat:](https:\u002F\u002Fgithub.com\u002FQianfengY\u002FUniRain) [:book:](.\u002Fcitations\u002Fyang2026unirain.txt)\n\n\u003Ca name=\"fine\" \u002F>\n\n### Fine-grained\n\n+ [FD2: A Dedicated Framework for Fine-Grained Dataset Distillation](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.25144) (Hongxu Ma & Guang Li et al., 2026) [:book:](.\u002Fcitations\u002Fma2026fd2.txt)\n\n## Media Coverage\n+ [Beginning of Awesome Dataset Distillation](https:\u002F\u002Ftwitter.com\u002FTongzhouWang\u002Fstatus\u002F1560043815204970497?cxt=HHwWgoCz9bPlsaYrAAAA)\n+ [Most Popular AI Research Aug 2022](https:\u002F\u002Fwww.libhunt.com\u002Fposts\u002F874974-d-most-popular-ai-research-aug-2022-ranked-based-on-github-stars)\n+ [一个项目帮你了解数据集蒸馏Dataset Distillation](https:\u002F\u002Fwww.jiqizhixin.com\u002Farticles\u002F2022-10-11-22)\n+ [浓缩就是精华：用大一统视角看待数据集蒸馏](https:\u002F\u002Fmp.weixin.qq.com\u002Fs\u002F__IjS0_FMpu35X9cNhNhPg)\n\n## Star History\n[![Star History Chart](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FGuang000_Awesome-Dataset-Distillation_readme_127a773fa9fc.png)](https:\u002F\u002Fstar-history.com\u002F#Guang000\u002FAwesome-Dataset-Distillation&Date)\n\n## Citing Awesome Dataset Distillation\nIf you find this project useful for your research, please use the following BibTeX entry.\n```\n@misc{li2022awesome,\n  author={Li, Guang and Zhao, Bo and Wang, Tongzhou},\n  title={Awesome Dataset Distillation},\n  howpublished={\\url{https:\u002F\u002Fgithub.com\u002FGuang000\u002FAwesome-Dataset-Distillation}},\n  year={2022}\n}\n```\n\n## Acknowledgments\nWe would like to express our heartfelt thanks to [Nikolaos Tsilivis](https:\u002F\u002Fgithub.com\u002FTsili42), [Wei Jin](https:\u002F\u002Fgithub.com\u002FChandlerBang), [Yongchao Zhou](https:\u002F\u002Fgithub.com\u002Fyongchao97), [Noveen Sachdeva](https:\u002F\u002Fgithub.com\u002Fnoveens), [Can Chen](https:\u002F\u002Fgithub.com\u002FGGchen1997), [Guangxiang Zhao](https:\u002F\u002Fgithub.com\u002Fzhaoguangxiang), [Shiye Lei](https:\u002F\u002Fgithub.com\u002FLeavesLei), [Xinchao Wang](https:\u002F\u002Fsites.google.com\u002Fsite\u002Fsitexinchaowang\u002F), [Dmitry Medvedev](https:\u002F\u002Fgithub.com\u002Fdm-medvedev), [Seungjae Shin](https:\u002F\u002Fgithub.com\u002FSJShin-AI), [Jiawei Du](https:\u002F\u002Fgithub.com\u002FAngusDujw), [Yidi Jiang](https:\u002F\u002Fgithub.com\u002FJiang-Yidi), [Xindi Wu](https:\u002F\u002Fgithub.com\u002FXindiWu), [Guangyi Liu](https:\u002F\u002Fgithub.com\u002Flgy0404), [Yilun Liu](https:\u002F\u002Fgithub.com\u002Fsuperallen13), [Kai Wang](https:\u002F\u002Fgithub.com\u002Fkaiwang960112), [Yue Xu](https:\u002F\u002Fgithub.com\u002Fsilicx), [Anjia Cao](https:\u002F\u002Fgithub.com\u002FCAOANJIA), [Jianyang Gu](https:\u002F\u002Fgithub.com\u002Fvimar-gu), [Yuanzhen Feng](https:\u002F\u002Fgithub.com\u002Ffengyzpku), [Peng Sun](https:\u002F\u002Fgithub.com\u002Fsp12138), [Ahmad Sajedi](https:\u002F\u002Fgithub.com\u002FAhmadSajedii), [Zhihao Sui](https:\u002F\u002Fgithub.com\u002Fsuizhihao), [Ziyu Wang](https:\u002F\u002Fgithub.com\u002Fyuz1wan), [Haoyang Liu](https:\u002F\u002Fgithub.com\u002FLiu-Hy), [Eduardo Montesuma](https:\u002F\u002Fgithub.com\u002Feddardd), [Shengbo Gong](https:\u002F\u002Fgithub.com\u002Frockcor), [Zheng Zhou](https:\u002F\u002Fgithub.com\u002Fzhouzhengqd), [Zhenghao Zhao](https:\u002F\u002Fgithub.com\u002Fichbill), [Duo Su](https:\u002F\u002Fgithub.com\u002Fsuduo94), [Tianhang Zheng](https:\u002F\u002Fgithub.com\u002Ftianzheng4), [Shijie Ma](https:\u002F\u002Fgithub.com\u002Fmashijie1028), [Wei Wei](https:\u002F\u002Fgithub.com\u002FWeiWeic6222848), [Yantai Yang](https:\u002F\u002Fgithub.com\u002FHiter-Q), [Shaobo Wang](https:\u002F\u002Fgithub.com\u002Fgszfwsb), [Xinhao Zhong](https:\u002F\u002Fgithub.com\u002Fndhg1213), [Zhiqiang Shen](https:\u002F\u002Fgithub.com\u002Fszq0214), [Cong Cong](https:\u002F\u002Fgithub.com\u002Fthomascong121), [Chun-Yin Huang](https:\u002F\u002Fgithub.com\u002Fchunyinhuang), [Dai Liu](https:\u002F\u002Fgithub.com\u002FNiaLiu), [Ruonan Yu](https:\u002F\u002Fgithub.com\u002FLexie-YU), [William Holland](https:\u002F\u002Fgithub.com\u002Frayneholland), [Saksham Singh Kushwaha](https:\u002F\u002Fgithub.com\u002Fsakshamsingh1), [Ping Liu](https:\u002F\u002Fgithub.com\u002Fpinglmlcv), [Wenliang Zhong](https:\u002F\u002Fgithub.com\u002FZhong0x29a), [Ning Li](https:\u002F\u002Fgithub.com\u002FNing9319), [Guochen Yan](https:\u002F\u002Fgithub.com\u002FYouth-49), [Saumyaranjan Mohanty](https:\u002F\u002Fgithub.com\u002Farareddy), and [Taehyung Kwon](https:\u002F\u002Fgithub.com\u002Fkbrother) for their valuable suggestions and contributions.\n\nThe [Homepage](https:\u002F\u002Fguang000.github.io\u002FAwesome-Dataset-Distillation\u002F) of Awesome Dataset Distillation was designed by [Longzhen Li](https:\u002F\u002Fgithub.com\u002FLOVELESSG) and maintained by [Mingzhuo Li](https:\u002F\u002Fgithub.com\u002FSumomoTaku).\n","# 令人惊叹的数据集蒸馏\n\n[![Awesome](https:\u002F\u002Fcdn.rawgit.com\u002Fsindresorhus\u002Fawesome\u002Fd7305f38d29fed78fa85652e3a63e154dd8e8829\u002Fmedia\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fsindresorhus\u002Fawesome)\n\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FContributions-Welcome-278ea5\" alt=\"Contrib\"\u002F> \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FNumber%20of%20Items-321-FF6F00\" alt=\"PaperNum\"\u002F> ![Stars](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002FGuang000\u002FAwesome-Dataset-Distillation?color=yellow&label=Stars) ![Forks](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fforks\u002FGuang000\u002FAwesome-Dataset-Distillation?color=green&label=Forks)\n\n**令人惊叹的数据集蒸馏** 提供了关于数据集蒸馏领域最全面、最详尽的信息。\n\n**数据集蒸馏** 是指合成一个小型数据集，使得在其上训练的模型能够在原始大型数据集上达到高性能的任务。数据集蒸馏算法以待蒸馏的大型真实数据集（训练集）作为**输入**，并**输出**一个小型的合成蒸馏数据集。该蒸馏数据集通过在独立的真实数据集（验证\u002F测试集）上评估基于此蒸馏数据集训练的模型来进行评价。一个好的小型蒸馏数据集不仅有助于理解数据集本身，还具有多种应用（例如持续学习、隐私保护、神经架构搜索等）。这一任务最早由论文[*Dataset Distillation* [Tongzhou Wang et al., '18]](https:\u002F\u002Fwww.tongzhouwang.info\u002Fdataset_distillation\u002F)提出，并提出了一种利用优化步骤中的反向传播来实现的算法。随后，在论文[*Medical Dataset Distillation* [Guang Li et al., '19]](https:\u002F\u002Farxiv.org\u002Fabs\u002F2104.02857)中，该任务首次被扩展到真实世界的数据集，并探讨了数据集蒸馏在隐私保护方面的可能性。而在论文[*Dataset Condensation* [Bo Zhao et al., '20]](https:\u002F\u002Farxiv.org\u002Fabs\u002F2006.05929)中，梯度匹配方法首次被引入，极大地推动了数据集蒸馏领域的发展。\n\n近年来（2022年至今），数据集蒸馏在研究界受到了越来越多的关注，许多机构和实验室都在积极开展相关研究。每年发表的相关论文数量也在不断增加。这些出色的研究不断改进数据集蒸馏技术，并探索其各种变体和应用场景。\n\n**本项目由 [Guang Li](https:\u002F\u002Fwww-lmd.ist.hokudai.ac.jp\u002Fmember\u002Fguang-li\u002F)、[Bo Zhao](https:\u002F\u002Fwww.bozhao.me\u002F) 和 [Tongzhou Wang](https:\u002F\u002Fwww.tongzhouwang.info\u002F) 共同策划和维护。**\n\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FGuang000_Awesome-Dataset-Distillation_readme_e5c26c8ba5f0.jpg\" width=\"20%\"\u002F>\n\n#### [如何提交拉取请求？](.\u002FCONTRIBUTING.md)\n\n+ :globe_with_meridians: 项目页面\n+ :octocat: 代码\n+ :book: `bibtex`\n\n## 最新动态\n+ [2026年4月2日] [面向数据集蒸馏的学习能力引导扩散](https:\u002F\u002Farxiv.org\u002Fabs\u002F2604.00519)（Jeffrey A. Chan-Santiago 等人，CVPR 2026）[:globe_with_meridians:](https:\u002F\u002Fjachansantiago.com\u002Flearnability-guided-distillation\u002F) [:book:](.\u002Fcitations\u002Fchansantiago2026learnability.txt)\n+ [2026年3月26日] [FD2：用于细粒度数据集蒸馏的专用框架](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.25144)（Hongxu Ma & Guang Li 等人，2026年）[:book:](.\u002Fcitations\u002Fma2026fd2.txt)\n+ [2026年3月26日] [DIET：为推荐系统持续蒸馏数据集的学习方法](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.24958)（Jiaqing Zhang 等人，2026年）[:book:](.\u002Fcitations\u002Fzhang2026diet.txt)\n+ [2026年3月26日] [数据集蒸馏高效编码非线性任务基于梯度学习的低维表示](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.14830)（Yuri Kinoshita 等人，2026年）[:book:](.\u002Fcitations\u002Fkinoshita2026lowdim.txt)\n+ [2026年3月26日] [ShapeCond：用于时间序列分类的快速形状子串引导数据集凝结](https:\u002F\u002Farxiv.org\u002Fabs\u002F2602.09008)（Sijia Peng 等人，2026年）[:octocat:](https:\u002F\u002Fgithub.com\u002Flunaaa95\u002FShapeCond) [:book:](.\u002Fcitations\u002Fpeng2026shapecond.txt)\n+ [2026年3月26日] [PRISM：通过解耦架构先验实现数据集蒸馏的多样化](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.09905)（Brian B. Moser 等人，TMLR 2026）[:octocat:](https:\u002F\u002Fgithub.com\u002FBrian-Moser\u002Fprism) [:book:](.\u002Fcitations\u002Fmoser2026prism.txt)\n+ [2026年3月23日] [IMS3：打破扩散式数据集蒸馏中的分布聚合](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.13960)（Chenru Wang & Yunyi Chen 等人，CVPR 2026）[:book:](.\u002Fcitations\u002Fwang2026ims3.txt)\n+ [2026年3月23日] [EVLF：用于生成式数据集蒸馏的早期视觉-语言融合](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.07476)（Wenqi Cai 等人，CVPR 2026）[:globe_with_meridians:](https:\u002F\u002Fwenqi-cai297.github.io\u002Fearlyfusion-HP\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fwenqi-cai297\u002Fearlyfusion-for-dd\u002F) [:book:](.\u002Fcitations\u002Fcai2026evlf.txt)\n+ [2026年3月23日] [HIERAMP：用于生成式数据集蒸馏的粗细结合自回归放大](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.06932)（Lin Zhao & Xinru Jiang 等人，CVPR 2026）[:octocat:](https:\u002F\u002Fgithub.com\u002FOshikaka\u002FHIERAMP) [:book:](.\u002Fcitations\u002Fzhao2026hieramp.txt)\n+ [2026年3月6日] [UniRain：基于RAG的数据集蒸馏与多目标重加权优化的统一图像去雨](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.03967)（Qianfeng Yang 等人，CVPR 2026）[:octocat:](https:\u002F\u002Fgithub.com\u002FQianfengY\u002FUniRain) [:book:](.\u002Fcitations\u002Fyang2026unirain.txt)\n+ [2026年3月6日] [固定锚点还不够：用于数据集蒸馏的动态检索与持久同调](https:\u002F\u002Farxiv.org\u002Fabs\u002F2602.24144)（Muquan Li 等人，CVPR 2026）[:book:](.\u002Fcitations\u002Fli2026reta.txt)\n+ [2026年3月6日] [ManifoldGD：用于扩散式数据集蒸馏的免训练分层流形指导](https:\u002F\u002Farxiv.org\u002Fabs\u002F2602.23295)（Ayush Roy 等人，CVPR 2026）[:octocat:](https:\u002F\u002Fgithub.com\u002FAyushRoy2001\u002FManifoldGD) [:book:](.\u002Fcitations\u002Froy2026manifold.txt)\n+ [2026年3月6日] [PRISM：用于稀疏运动视频的数据集凝结，采用渐进式精炼与插入](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.22564)（Jaehyun Choi 等人，CVPR 2026）[:book:](.\u002Fcitations\u002Fchoi2026prism.txt)\n\n## 目录\n- [主条目](#main)\n  - [早期工作](#early-work)\n  - [梯度\u002F轨迹匹配代理目标函数](#gradient-objective)\n  - [分布\u002F特征匹配代理目标函数](#feature-objective)\n  - [基于核的蒸馏](#kernel)\n  - [蒸馏数据集参数化](#parametrization)\n  - [生成式蒸馏](#generative)\n  - [更优的优化方法](#optimization)\n  - [更深入的理解](#understanding)\n  - [标签蒸馏](#label)\n  - [数据集量化](#quant)\n  - [解耦蒸馏](#decouple)\n  - [多模态蒸馏](#multi)\n  - [自监督蒸馏](#self)\n  - [基准测试](#benchmark)\n  - [综述](#survey)\n  - [博士论文](#thesis)\n  - [研讨会](#workshop)\n  - [挑战赛](#challenge)\n- [应用](#applications)\n  - [持续学习](#continual)\n  - [隐私保护](#privacy)\n  - [医疗](#medical)\n  - [联邦学习](#fed)\n  - [图神经网络](#gnn)\n  - [神经架构搜索](#nas)\n  - [时尚、艺术和设计](#fashion)\n  - [推荐系统](#rec)\n  - [黑盒优化](#blackbox)\n  - [鲁棒性](#robustness)\n  - [公平性](#fairness)\n  - [文本](#text)\n  - [视频](#video)\n  - [表格数据](#tabular)\n  - [检索](#retrieval)\n  - [领域适应](#domain)\n  - [超分辨率](#super)\n  - [时间序列](#time)\n  - [语音](#speech)\n  - [机器去学习](#unlearning)\n  - [强化学习](#rl)\n  - [长尾分布](#long)\n  - [噪声标签学习](#noisy)\n  - [目标检测](#detection)\n  - [点云](#point)\n  - [通用蒸馏](#uni)\n  - [脉冲神经网络](#snn)\n  - [脑电图](#eeg)\n  - [金融](#finance)\n  - [音乐](#music)\n  - [遥感](#rs)\n  - [去雨](#dr)\n  - [细粒度识别](#fine)\n\u003Ca name=\"main\" \u002F>\n\n## 主条目\n+ [数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F1811.10959)（Tongzhou Wang 等，2018年）[:globe_with_meridians:](https:\u002F\u002Fssnl.github.io\u002Fdataset_distillation\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002FSsnL\u002Fdataset-distillation) [:book:](.\u002Fcitations\u002Fwang2018datasetdistillation.txt)\n\n\u003Ca name=\"early-work\" \u002F>\n\n### 早期工作\n+ [基于梯度的超参数优化：通过可逆学习实现](https:\u002F\u002Farxiv.org\u002Fabs\u002F1502.03492)（Dougal Maclaurin 等，ICML 2015）[:octocat:](https:\u002F\u002Fgithub.com\u002FHIPS\u002Fhypergrad) [:book:](.\u002Fcitations\u002Fmaclaurin2015gradient.txt)\n\n\u003Ca name=\"gradient-objective\" \u002F>\n\n### 基于梯度\u002F轨迹匹配的代理目标函数\n+ [通过梯度匹配进行数据集浓缩](https:\u002F\u002Farxiv.org\u002Fabs\u002F2006.05929)（Bo Zhao 等，ICLR 2021）[:octocat:](https:\u002F\u002Fgithub.com\u002FVICO-UoE\u002FDatasetCondensation) [:book:](.\u002Fcitations\u002Fzhao2021datasetcondensation.txt)\n+ [通过可微分的双生增强进行数据集浓缩](https:\u002F\u002Farxiv.org\u002Fabs\u002F2102.08259)（Bo Zhao 等，ICML 2021）[:octocat:](https:\u002F\u002Fgithub.com\u002FVICO-UoE\u002FDatasetCondensation) [:book:](.\u002Fcitations\u002Fzhao2021differentiatble.txt)\n+ [通过匹配训练轨迹进行数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2203.11932)（George Cazenavette 等，CVPR 2022）[:globe_with_meridians:](https:\u002F\u002Fgeorgecazenavette.github.io\u002Fmtt-distillation\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fgeorgecazenavette\u002Fmtt-distillation) [:book:](.\u002Fcitations\u002Fcazenavette2022dataset.txt)\n+ [利用对比信号进行数据集浓缩](https:\u002F\u002Farxiv.org\u002Fabs\u002F2202.02916)（Saehyung Lee 等，ICML 2022）[:octocat:](https:\u002F\u002Fgithub.com\u002Fsaehyung-lee\u002Fdcc) [:book:](.\u002Fcitations\u002Flee2022dataset.txt)\n+ [通过损失曲率匹配进行数据集选择与浓缩](https:\u002F\u002Farxiv.org\u002Fabs\u002F2303.04449)（Seungjae Shin & Heesun Bae 等，AISTATS 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002FSJShin-AI\u002FLCMat) [:book:](.\u002Fcitations\u002Fshin2023lcmat.txt)\n+ [通过最小化累积轨迹误差提升数据集蒸馏效果](https:\u002F\u002Farxiv.org\u002Fabs\u002F2211.11004)（Jiawei Du & Yidi Jiang 等，CVPR 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002FAngusDujw\u002FFTD-distillation) [:book:](.\u002Fcitations\u002Fdu2023minimizing.txt)\n+ [以恒定内存规模将数据集蒸馏扩展至 ImageNet-1K](https:\u002F\u002Farxiv.org\u002Fabs\u002F2211.10586)（Justin Cui 等，ICML 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002Fjustincui03\u002Ftesla) [:book:](.\u002Fcitations\u002Fcui2022scaling.txt) \n+ [用于数据集蒸馏的顺序子集匹配](https:\u002F\u002Farxiv.org\u002Fabs\u002F2311.01570)（Jiawei Du 等，NeurIPS 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002Fshqii1j\u002Fseqmatch) [:book:](.\u002Fcitations\u002Fdu2023seqmatch.txt)\n+ [通过难度对齐的轨迹匹配实现无损数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.05773)（Ziyao Guo & Kai Wang 等，ICLR 2024）[:globe_with_meridians:](https:\u002F\u002Fgzyaftermath.github.io\u002FDATM\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002FGzyAftermath\u002FDATM) [:book:](.\u002Fcitations\u002Fguo2024datm.txt)\n+ [SelMatch：通过基于选择的初始化和部分更新，结合轨迹匹配有效扩展数据集蒸馏规模](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.18561)（Yongmin Lee 等，ICML 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002FYongalls\u002FSelMatch) [:book:](.\u002Fcitations\u002Flee2024selmatch.txt)\n+ [通过自动训练轨迹进行数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2407.14245)（Dai Liu 等，ECCV 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002FNiaLiu\u002FATT) [:book:](.\u002Fcitations\u002Fliu2024att.txt)\n+ [用于数据集蒸馏的神经谱分解](https:\u002F\u002Farxiv.org\u002Fabs\u002F2408.16236)（Shaolei Yang 等，ECCV 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Fslyang2021\u002FNSD) [:book:](.\u002Fcitations\u002Fyang2024nsd.txt)\n+ [在数据集蒸馏中优先考虑对齐](https:\u002F\u002Farxiv.org\u002Fabs\u002F2408.03360)（Zekai Li & Ziyao Guo 等，2024年）[:octocat:](https:\u002F\u002Fgithub.com\u002FNUS-HPC-AI-Lab\u002FPAD) [:book:](.\u002Fcitations\u002Fli2024pad.txt)\n+ [迈向稳定且节省存储空间的数据集蒸馏：匹配凸化轨迹](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.19827)（Wenliang Zhong 等，CVPR 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002FZhong0x29a\u002FMCT) [:book:](.\u002Fcitations\u002Fzhong2025mct.txt)\n+ [在复杂场景下强调判别特征以进行数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2410.17193)（Kai Wang & Zekai Li 等，CVPR 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002FNUS-HPC-AI-Lab\u002FEDF) [:book:](.\u002Fcitations\u002Fwang2025edf.txt)\n\n\u003Ca name=\"feature-objective\" \u002F>\n\n### 分布\u002F特征匹配替代目标\n+ [CAFE：通过对齐特征学习数据集压缩](https:\u002F\u002Farxiv.org\u002Fabs\u002F2203.01531)（Kai Wang & Bo Zhao 等，CVPR 2022）[:octocat:](https:\u002F\u002Fgithub.com\u002Fkaiwang960112\u002Fcafe) [:book:](.\u002Fcitations\u002Fwang2022cafe.txt)\n+ [基于分布匹配的数据集压缩](https:\u002F\u002Farxiv.org\u002Fabs\u002F2110.04181)（Bo Zhao 等，WACV 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002FVICO-UoE\u002FDatasetCondensation) [:book:](.\u002Fcitations\u002Fzhao2023distribution.txt)\n+ [用于数据集压缩的改进分布匹配](https:\u002F\u002Farxiv.org\u002Fabs\u002F2307.09742)（Ganlong Zhao 等，CVPR 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002Fuitrbn\u002FIDM) [:book:](.\u002Fcitations\u002Fzhao2023idm.txt)\n+ [DataDAM：基于注意力匹配的高效数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.00093)（Ahmad Sajedi & Samir Khaki 等，ICCV 2023）[:globe_with_meridians:](https:\u002F\u002Fdatadistillation.github.io\u002FDataDAM\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002FDataDistillation\u002FDataDAM) [:book:](.\u002Fcitations\u002Fsajedi2023datadam.txt)\n+ [M3D：通过最小化最大均值差异进行数据集压缩](https:\u002F\u002Farxiv.org\u002Fabs\u002F2312.15927)（Hansong Zhang & Shikun Li 等，AAAI 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002FHansong-Zhang\u002FM3D) [:book:](.\u002Fcitations\u002Fzhang2024m3d.txt)\n+ [在数据集蒸馏中利用样本间与特征间的关联](https:\u002F\u002Farxiv.org\u002Fabs\u002F2404.00563)（Wenxiao Deng 等，CVPR 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002FVincenDen\u002FIID) [:book:](.\u002Fcitations\u002Fdeng2024iid.txt)\n+ [基于潜在分位数匹配的数据集压缩](https:\u002F\u002Fopenaccess.thecvf.com\u002Fcontent\u002FCVPR2024W\u002FDDCV\u002Fhtml\u002FWei_Dataset_Condensation_with_Latent_Quantile_Matching_CVPRW_2024_paper.html)（Wei Wei 等，CVPR 2024研讨会）[:book:](.\u002Fcitations\u002Fwei2024lqm.txt)\n+ [DANCE：用于数据集压缩的双视角分布对齐](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.01063)（Hansong Zhang 等，IJCAI 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002FHansong-Zhang\u002FDANCE) [:book:](.\u002Fcitations\u002Fzhang2024dance.txt)\n+ [面向数据集蒸馏的多样化语义分布匹配](https:\u002F\u002Fdl.acm.org\u002Fdoi\u002F10.1145\u002F3664647.3680900)（Hongcheng Li 等，MM 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002FLi-Hongcheng\u002FDSDM) [:book:](.\u002Fcitations\u002Fli2024dsdm.txt)\n+ [基于神经特征函数的数据集蒸馏：一种极小极大视角](https:\u002F\u002Farxiv.org\u002Fabs\u002F2502.20653)（Shaobo Wang 等，CVPR 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002Fgszfwsb\u002FNCFM) [:book:](.\u002Fcitations\u002Fwang2025ncfm.txt)\n+ [OPTICAL：利用最优传输进行数据集蒸馏中的贡献分配](https:\u002F\u002Fopenaccess.thecvf.com\u002Fcontent\u002FCVPR2025\u002Fhtml\u002FCui_OPTICAL_Leveraging_Optimal_Transport_for_Contribution_Allocation_in_Dataset_Distillation_CVPR_2025_paper.html)（Xiao Cui 等，CVPR 2025）[:book:](.\u002Fcitations\u002Fcui2025optical.txt)\n+ [基于Wasserstein度量的数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2311.18531)（Haoyang Liu 等，ICCV 2025）[:globe_with_meridians:](https:\u002F\u002Fliu-hy.github.io\u002FWMDD\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002FLiu-Hy\u002FWMDD) [:book:](.\u002Fcitations\u002Fliu2025wasserstein.txt)\n+ [面向数据集蒸馏的多样性增强分布对齐](https:\u002F\u002Fopenaccess.thecvf.com\u002Fcontent\u002FICCV2025\u002Fhtml\u002FLi_Diversity-Enhanced_Distribution_Alignment_for_Dataset_Distillation_ICCV_2025_paper.html)（Hongcheng Li 等，ICCV 2025）[:book:](.\u002Fcitations\u002Fli2025deda.txt)\n+ [双曲空间中的数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.24623)（Wenyuan Li & Guang Li 等，NeurIPS 2025）[:globe_with_meridians:](https:\u002F\u002Fguang000.github.io\u002FHDD-Webpage\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002FGuang000\u002FHDD) [:book:](.\u002Fcitations\u002Fli2025hdd.txt)\n+ [TGDD：基于轨迹引导且分布均衡的数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2512.02469)（Fengli Ran 等，AAAI 2026）[:octocat:](https:\u002F\u002Fgithub.com\u002FFlyFinley\u002FTGDD) [:book:](.\u002Fcitations\u002Fran2026tgdd.txt)\n\n\u003Ca name=\"kernel\" \u002F>\n\n### 基于核的蒸馏\n+ [基于核岭回归的数据集元学习](https:\u002F\u002Farxiv.org\u002Fabs\u002F2011.00050)（Timothy Nguyen 等，ICLR 2021）[:octocat:](https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fneural-tangents) [:book:](.\u002Fcitations\u002Fnguyen2021kip.txt)\n+ [使用无限宽卷积网络进行数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2107.13034)（Timothy Nguyen 等，NeurIPS 2021）[:octocat:](https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fneural-tangents) [:book:](.\u002Fcitations\u002Fnguyen2021kipimprovedresults.txt)\n+ [利用神经特征回归进行数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2206.00719)（Yongchao Zhou 等，NeurIPS 2022）[:globe_with_meridians:](https:\u002F\u002Fsites.google.com\u002Fview\u002Ffrepo) [:octocat:](https:\u002F\u002Fgithub.com\u002Fyongchao97\u002FFRePo) [:book:](.\u002Fcitations\u002Fzhou2022dataset.txt)\n+ [使用随机特征近似进行高效数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2210.12067)（Noel Loo 等，NeurIPS 2022）[:octocat:](https:\u002F\u002Fgithub.com\u002Fyolky\u002FRFAD) [:book:](.\u002Fcitations\u002Floo2022efficient.txt)\n+ [通过凸化隐式梯度进行数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2302.06755)（Noel Loo 等，ICML 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002Fyolky\u002FRCIG) [:book:](.\u002Fcitations\u002Floo2023dataset.txt)\n+ [针对核岭回归的可证明且高效的数据集蒸馏](https:\u002F\u002Fopenreview.net\u002Fforum?id=WI2VpcBdnd)（Yilan Chen 等，NeurIPS 2024）[:book:](.\u002Fcitations\u002Fchen2024krr.txt)\n\n\u003Ca name=\"parametrization\" \u002F>\n\n### 数据集蒸馏参数化\n+ [通过高效合成数据参数化进行数据集凝缩](https:\u002F\u002Farxiv.org\u002Fabs\u002F2205.14959)（Jang-Hyun Kim 等，ICML 2022）[:octocat:](https:\u002F\u002Fgithub.com\u002Fsnu-mllab\u002Fefficient-dataset-condensation) [:book:](.\u002Fcitations\u002Fkim2022dataset.txt)\n+ [铭记过去：将数据集蒸馏为神经网络的可寻址记忆](https:\u002F\u002Farxiv.org\u002Fabs\u002F2206.02916)（Zhiwei Deng 等，NeurIPS 2022）[:octocat:](https:\u002F\u002Fgithub.com\u002Fprincetonvisualai\u002FRememberThePast-DatasetDistillation) [:book:](.\u002Fcitations\u002Fdeng2022remember.txt)\n+ [关于贝叶斯伪核心集的散度度量](https:\u002F\u002Farxiv.org\u002Fabs\u002F2210.06205)（Balhae Kim 等，NeurIPS 2022）[:octocat:](https:\u002F\u002Fgithub.com\u002Fbalhaekim\u002Fbpc-divergences) [:book:](.\u002Fcitations\u002Fkim2022divergence.txt)\n+ [通过因子分解进行数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2210.16774)（Songhua Liu 等，NeurIPS 2022）[:octocat:](https:\u002F\u002Fgithub.com\u002FHuage001\u002FDatasetFactorization) [:book:](.\u002Fcitations\u002Fliu2022dataset.txt)\n+ [PRANC：用于压缩深度模型的伪随机网络](https:\u002F\u002Farxiv.org\u002Fabs\u002F2206.08464)（Parsa Nooralinejad 等，2022）[:octocat:](https:\u002F\u002Fgithub.com\u002FUCDvision\u002FPRANC) [:book:](.\u002Fcitations\u002Fnooralinejad2022pranc.txt)\n+ [基于潜在空间知识因子分解与共享的数据集凝缩](https:\u002F\u002Farxiv.org\u002Fabs\u002F2208.10494)（Hae Beom Lee & Dong Bok Lee 等，2022）[:book:](.\u002Fcitations\u002Flee2022kfs.txt)\n+ [可瘦身数据集凝缩](https:\u002F\u002Fopenaccess.thecvf.com\u002Fcontent\u002FCVPR2023\u002Fhtml\u002FLiu_Slimmable_Dataset_Condensation_CVPR_2023_paper.html)（Songhua Liu 等，CVPR 2023）[:book:](.\u002Fcitations\u002Fliu2023slimmable.txt)\n+ [基于迁移预训练的少样本数据集蒸馏](https:\u002F\u002Fopenaccess.thecvf.com\u002Fcontent\u002FICCV2023\u002Fhtml\u002FLiu_Few-Shot_Dataset_Distillation_via_Translative_Pre-Training_ICCV_2023_paper.html)（Songhua Liu 等，ICCV 2023）[:book:](.\u002Fcitations\u002Fliu2023fewshot.txt)\n+ [MGDD：用于快速数据集蒸馏的元生成器](https:\u002F\u002Fopenreview.net\u002Fforum?id=D9CMRR5Lof)（Songhua Liu 等，NeurIPS 2023）[:book:](.\u002Fcitations\u002Fliu2023mgdd.txt)\n+ [用于表征性数据集蒸馏的稀疏参数化](https:\u002F\u002Fopenreview.net\u002Fforum?id=ZIfhYAE2xg)（Xing Wei & Anjia Cao 等，NeurIPS 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002FMIV-XJTU\u002FSPEED) [:book:](.\u002Fcitations\u002Fwei2023sparse.txt)\n+ [基于频域的数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2311.08819)（Donghyeok Shin & Seungjae Shin 等，NeurIPS 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002Fsdh0818\u002FFreD) [:book:](.\u002Fcitations\u002Fshin2023fred.txt)\n+ [利用层次化特征共享实现高效数据集凝缩](https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.07506)（Haizhong Zheng 等，ECCV 2024）[:book:](.\u002Fcitations\u002Fzheng2024hmn.txt)\n+ [FYI：为数据集蒸馏翻转你的图像](https:\u002F\u002Farxiv.org\u002Fabs\u002F2407.08113)（Byunggwan Son 等，ECCV 2024）[:globe_with_meridians:](https:\u002F\u002Fcvlab.yonsei.ac.kr\u002Fprojects\u002FFYI\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fcvlab-yonsei\u002FFYI) [:book:](.\u002Fcitations\u002Fson2024fyi.txt)\n+ [数据集蒸馏中的色彩导向冗余去除](https:\u002F\u002Farxiv.org\u002Fabs\u002F2411.11329)（Bowen Yuan 等，NeurIPS 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002FKeViNYuAn0314\u002FAutoPalette) [:book:](.\u002Fcitations\u002Fyuan2024color.txt)\n+ [将数据集蒸馏为神经场](https:\u002F\u002Farxiv.org\u002Fabs\u002F2503.04835)（Donghyeok Shin 等，ICLR 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002Faailab-kaist\u002FDDiF) [:book:](.\u002Fcitations\u002Fshin2025ddif.txt)\n+ [作为数据压缩的数据集蒸馏：速率-效用视角](https:\u002F\u002Farxiv.org\u002Fabs\u002F2507.17221)（Youneng Bao & Yiping Liu 等，ICCV 2025）[:globe_with_meridians:](https:\u002F\u002Fnouise.github.io\u002FDD-RUO\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fnouise\u002FDD-RUO) [:book:](.\u002Fcitations\u002Fbao2025ruo.txt)\n+ [超越像素：基于稀疏高斯表示的高效数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2509.26219)（Chenyang Jiang 等，2025）[:octocat:](https:\u002F\u002Fgithub.com\u002Fj-cyoung\u002FGSDatasetDistillation) [:book:](.\u002Fcitations\u002Fjiang2025gsdd.txt)\n+ [用于高效数据集凝缩的训练后量化](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.13346)（Linh-Tam Tran 等，AAAI 2026）[:book:](.\u002Fcitations\u002Ftran2026ptqdc.txt)\n\n\u003Ca name=\"generative\" \u002F>\n\n### 生成式蒸馏\n#### GAN\n+ [使用 GAN 合成信息丰富的训练样本](https:\u002F\u002Farxiv.org\u002Fabs\u002F2204.07513)（Bo Zhao 等，NeurIPS 2022 工作坊）[:octocat:](https:\u002F\u002Fgithub.com\u002Fvico-uoe\u002Fit-gan) [:book:](.\u002Fcitations\u002Fzhao2022synthesizing.txt)\n+ [通过深度生成先验泛化数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2305.01649)（George Cazenavette 等，CVPR 2023）[:globe_with_meridians:](https:\u002F\u002Fgeorgecazenavette.github.io\u002Fglad\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fgeorgecazenavette\u002Fglad) [:book:](.\u002Fcitations\u002Fcazenavette2023glad.txt)\n+ [DiM：将数据集蒸馏为生成模型](https:\u002F\u002Farxiv.org\u002Fabs\u002F2303.04707)（Kai Wang & Jianyang Gu 等，2023）[:octocat:](https:\u002F\u002Fgithub.com\u002Fvimar-gu\u002FDiM) [:book:](.\u002Fcitations\u002Fwang2023dim.txt)\n+ [通过生成模型进行数据集凝缩](https:\u002F\u002Farxiv.org\u002Fabs\u002F2309.07698)（Junhao Zhang 等，2023）[:book:](.\u002Fcitations\u002Fzhang2023dc.txt)\n+ [生成式数据集蒸馏：平衡全局结构与局部细节](https:\u002F\u002Farxiv.org\u002Fabs\u002F2404.17732)（Longzhen Li & Guang Li 等，CVPR 2024 工作坊）[:book:](.\u002Fcitations\u002Fli2024generative.txt)\n+ [从数据到模型的蒸馏：数据高效学习框架](https:\u002F\u002Fwww.ecva.net\u002Fpapers\u002Feccv_2024\u002Fpapers_ECCV\u002Fhtml\u002F6020_ECCV_2024_paper.php)（Ahmad Sajedi & Samir Khaki 等，ECCV 2024）[:book:](.\u002Fcitations\u002Fsajedi2024data.txt)\n+ [基于自我知识蒸馏的生成式数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2501.04202)（Longzhen Li & Guang Li 等，ICASSP 2025）[:book:](.\u002Fcitations\u002Fli2025generative.txt)\n+ [层次化特征很重要：深入探索 GAN 先验以改进数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.05704)（Xinhao Zhong & Hao Fang 等，CVPR 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002Fndhg1213\u002FH-GLaD) [:book:](.\u002Fcitations\u002Fzhong2025hglad.txt)\n\n#### 扩散\n+ [通过极小极大扩散高效进行数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2311.15529) (Jianyang Gu 等，CVPR 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fvimar-gu\u002FMinimaxDiffusion) [:book:](.\u002Fcitations\u002Fgu2024efficient.txt)\n+ [D4M：基于解耦扩散模型的数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2407.15138) (Duo Su & Junjie Hou 等，CVPR 2024) [:globe_with_meridians:](https:\u002F\u002Fjunjie31.github.io\u002FD4M\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fsuduo94\u002FD4M) [:book:](.\u002Fcitations\u002Fsu2024d4m.txt)\n+ [基于扩散模型的生成式数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2408.08610) (Duo Su & Junjie Hou & Guang Li 等，ECCV 2024 Workshop) [:octocat:](https:\u002F\u002Fgithub.com\u002FGuang000\u002FGenerative-Dataset-Distillation-Based-on-Diffusion-Model) [:book:](.\u002Fcitations\u002Fsu2024diffusion.txt)\n+ [用于数据集蒸馏的影响引导扩散](https:\u002F\u002Fopenreview.net\u002Fforum?id=0whx8MhysK) (Mingyang Chen 等，ICLR 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002Fmchen725\u002FDD_IGD) [:book:](.\u002Fcitations\u002Fchen2025igd.txt)\n+ [利用高代表性驯服扩散进行数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.18399) (Lin Zhao 等，ICML 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002Flin-zhao-resoLve\u002FD3HR) [:book:](.\u002Fcitations\u002Fzhao2025d3hr.txt)\n+ [MGD3：使用扩散模型的模式引导数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.18963) (Jeffrey A. Chan-Santiago 等，ICML 2025) [:globe_with_meridians:](https:\u002F\u002Fjachansantiago.com\u002Fmode-guided-distillation\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fjachansantiago\u002Fmode_guidance\u002F) [:book:](.\u002Fcitations\u002Fchan-santiago2025mgd3.txt)\n+ [通过对抗引导课程采样增强基于扩散的数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2508.01264) (Lexiao Zou 等，ICME 2025) [:book:](.\u002Fcitations\u002Fzou2025acs.txt)\n+ [CaO2：修正基于扩散的数据集蒸馏中的不一致性](https:\u002F\u002Farxiv.org\u002Fabs\u002F2506.22637) (Haoxuan Wang 等，ICCV 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002FhatchetProject\u002FCaO2) [:book:](.\u002Fcitations\u002Fwang2025cao2.txt)\n+ [通过视觉-语言类别原型进行数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2506.23580) (Yawen Zou & Guang Li 等，ICCV 2025) [:globe_with_meridians:](https:\u002F\u002Fzou-yawen.github.io\u002FDD_via_vision-language)  [:octocat:](https:\u002F\u002Fgithub.com\u002Fzou-yawen\u002FDataset-Distillation-via-Vision-Language-Category-Prototype\u002F) [:book:](.\u002Fcitations\u002Fzou2025vlcp.txt)\n+ [具有难度引导采样的特定任务生成式数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2507.03331) (Mingzhuo Li & Guang Li 等，ICCV 2025 Workshop) [:octocat:](https:\u002F\u002Fgithub.com\u002FSumomoTaku\u002FDiffGuideSamp) [:book:](.\u002Fcitations\u002Fli2025diff.txt)\n+ [利用扩散模型解锁数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2403.03881) (Brian B. Moser & Federico Raue 等，NeurIPS 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002FBrian-Moser\u002Fprune_and_distill) [:book:](.\u002Fcitations\u002Fmoser2025ld3m.txt)\n+ [利用最优传输优化分布几何对齐以进行生成式数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2512.00308) (Xiao Cui 等，NeurIPS 2025) [:book:](.\u002Fcitations\u002Fcui2025ot.txt)\n+ [带有颜色补偿的数据集凝聚](https:\u002F\u002Farxiv.org\u002Fabs\u002F2508.01139) (Huyu Wu 等，TMLR 2025) [:globe_with_meridians:](https:\u002F\u002F528why.github.io\u002FDC3-Page\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002F528why\u002FDataset-Condensation-with-Color-Compensation) [:book:](.\u002Fcitations\u002Fwu2025dc3.txt)\n+ [扩散模型作为数据集蒸馏先验](https:\u002F\u002Farxiv.org\u002Fabs\u002F2510.17421) (Duo Su 等，ICLR 2026) [:book:](.\u002Fcitations\u002Fsu2026dap.txt)\n+ [CoDA：从文本到图像的扩散模型到无训练数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2512.03844) (Letian Zhou 等，ICLR 2026) [:octocat:](https:\u002F\u002Fgithub.com\u002Fzzzlt422\u002FCoDA) [:book:](.\u002Fcitations\u002Fzhou2026coda.txt)\n+ [ManifoldGD：用于基于扩散的数据集蒸馏的无训练分层流形指导](https:\u002F\u002Farxiv.org\u002Fabs\u002F2602.23295) (Ayush Roy 等，CVPR 2026) [:octocat:](https:\u002F\u002Fgithub.com\u002FAyushRoy2001\u002FManifoldGD) [:book:](.\u002Fcitations\u002Froy2026manifold.txt)\n+ [IMS3：打破基于扩散的数据集蒸馏中的分布聚合](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.13960) (Chenru Wang & Yunyi Chen 等，CVPR 2026) [:book:](.\u002Fcitations\u002Fwang2026ims3.txt)\n+ [EVLF：用于生成式数据集蒸馏的早期视觉-语言融合](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.07476) (Wenqi Cai 等，CVPR 2026) [:globe_with_meridians:](https:\u002F\u002Fwenqi-cai297.github.io\u002Fearlyfusion-HP\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fwenqi-cai297\u002Fearlyfusion-for-dd\u002F) [:book:](.\u002Fcitations\u002Fcai2026evlf.txt)\n+ [用于数据集蒸馏的学习能力引导扩散](https:\u002F\u002Farxiv.org\u002Fabs\u002F2604.00519) (Jeffrey A. Chan-Santiago 等，CVPR 2026) [:globe_with_meridians:](https:\u002F\u002Fjachansantiago.com\u002Flearnability-guided-distillation\u002F) [:book:](.\u002Fcitations\u002Fchansantiago2026learnability.txt)\n\n#### VAR\n+ [HIERAMP：用于生成式数据集蒸馏的由粗到细自回归放大](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.06932) (Lin Zhao & Xinru Jiang 等，CVPR 2026) [:octocat:](https:\u002F\u002Fgithub.com\u002FOshikaka\u002FHIERAMP) [:book:](.\u002Fcitations\u002Fzhao2026hieramp.txt)\n\n#### 流\n+ [路径引导流匹配用于数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2602.05616) (Xuhui Li 等，2026) [:book:](.\u002Fcitations\u002Fli2026flow.txt)\n\n\u003Ca name=\"optimization\" \u002F>\n\n### 更好的优化\n+ [通过模型增强加速数据蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2212.06152) (Lei Zhang & Jie Zhang 等，CVPR 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Fncsu-dk-lab\u002FAcc-DD) [:book:](.\u002Fcitations\u002Fzhang2023accelerating.txt)\n+ [DREAM：基于代表性匹配的高效数据蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2302.14416) (Yanqing Liu & Jianyang Gu & Kai Wang 等，ICCV 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Flyq312318224\u002FDREAM) [:book:](.\u002Fcitations\u002Fliu2023dream.txt)\n+ [只蒸馏一次：压缩数据集的两条规则](https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.14019) (Yang He 等，NeurIPS 2023) [:octocat:](https:\u002F\u002Fgithub.com\u002Fhe-y\u002Fyou-only-condense-once) [:book:](.\u002Fcitations\u002Fhe2023yoco.txt)\n+ [MIM4DD：用于数据蒸馏的互信息最大化](https:\u002F\u002Farxiv.org\u002Fabs\u002F2312.16627) (Yuzhang Shang 等，NeurIPS 2023) [:book:](.\u002Fcitations\u002Fshang2023mim4dd.txt)\n+ [预训练模型能否助力数据蒸馏？](https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.03295) (Yao Lu 等，2023年) [:octocat:](https:\u002F\u002Fgithub.com\u002Fyaolu-zjut\u002FDDInterpreter) [:book:](.\u002Fcitations\u002Flu2023pre.txt)\n+ [DREAM+：双向代表性匹配的高效数据蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.15052) (Yanqing Liu & Jianyang Gu & Kai Wang 等，2023年) [:octocat:](https:\u002F\u002Fgithub.com\u002Flyq312318224\u002FDREAM) [:book:](.\u002Fcitations\u002Fliu2023dream+.txt)\n+ [潜在空间中的数据蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2311.15547) (Yuxuan Duan 等，2023年) [:book:](.\u002Fcitations\u002Fduan2023latent.txt)\n+ [数据蒸馏就像伏特加一样：多次蒸馏以获得更佳品质](https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.06982) (Xuxi Chen & Yu Yang 等，ICLR 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002FVITA-Group\u002FProgressiveDD) [:book:](.\u002Fcitations\u002Fchen2024vodka.txt)\n+ [极其简单的数据蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2311.07025) (Yunzhen Feng 等，ICLR 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Ffengyzpku\u002FSimple_Dataset_Distillation) [:book:](.\u002Fcitations\u002Fyunzhen2024embarassingly.txt)\n+ [多尺度数据集压缩](https:\u002F\u002Farxiv.org\u002Fabs\u002F2403.06075) (Yang He 等，ICLR 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fhe-y\u002FMultisize-Dataset-Condensation) [:book:](.\u002Fcitations\u002Fhe2024mdc.txt)\n+ [具有领域偏移的大规模数据蒸馏](https:\u002F\u002Fopenreview.net\u002Fforum?id=0FWPKHMCSc) (Noel Loo & Alaa Maalouf 等，ICML 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fyolky\u002Fd3s_distillation) [:book:](.\u002Fcitations\u002Floo2024d3s.txt)\n+ [从海量矿石中提炼黄金：面向高效数据蒸馏的双层数据剪枝](https:\u002F\u002Farxiv.org\u002Fabs\u002F2305.18381) (Yue Xu 等，ECCV 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fsilicx\u002FGoldFromOres) [:book:](.\u002Fcitations\u002Fxu2024distill.txt)\n+ [基于异构模型的模型无关数据集压缩](https:\u002F\u002Farxiv.org\u002Fabs\u002F2409.14538) (Jun-Yeong Moon 等，ECCV 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fkhu-agi\u002Fhmdc) [:book:](.\u002Fcitations\u002Fmoon2024hmdc.txt)\n+ [泰迪：基于泰勒近似匹配的高效大规模数据蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2410.07579) (Ruonan Yu 等，ECCV 2024) [:book:](.\u002Fcitations\u002Fyu2024teddy.txt)\n+ [BACON：用于数据蒸馏的贝叶斯最优压缩框架](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.01112) (Zheng Zhou 等，2024年) [:octocat:](https:\u002F\u002Fgithub.com\u002Fzhouzhengqd\u002FBACON) [:book:](.\u002Fcitations\u002Fzhou2024bacon.txt)\n+ [超越特征相似性：基于类感知条件互信息的有效数据蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2412.09945) (Xinhao Zhong 等，ICLR 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002Fndhg1213\u002FCMIDD) [:book:](.\u002Fcitations\u002Fzhong2025cmi.txt)\n+ [高IPC数据蒸馏的课程式粗细选择](https:\u002F\u002Farxiv.org\u002Fabs\u002F2503.18872) (Yanda Chen & Gongwei Chen 等，CVPR 2025) [:octocat:](https:\u002F\u002Fgithub.com\u002FCYDaaa30\u002FCCFS) [:book:](.\u002Fcitations\u002Fchen2025ccfs.txt)\n+ [并非所有样本都应同等对待：迈向对数据蒸馏的理解与改进](https:\u002F\u002Farxiv.org\u002Fabs\u002F2408.12483) (Shaobo Wang 等，CVPR 2025研讨会) [:book:](.\u002Fcitations\u002Fwang2025samples.txt)\n+ [超越随机：数据蒸馏中的自动内循环优化](https:\u002F\u002Farxiv.org\u002Fabs\u002F2510.04838) (Muquan Li 等，NeurIPS 2025) [:book:](.\u002Fcitations\u002Fli2025bptt.txt)\n+ [作为前向最优量化的数据蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2501.07681) (Hongye Tan 等，ICLR 2026) [:book:](.\u002Fcitations\u002Ftan2026optimal.txt)\n\n\n\u003Ca name=\"understanding\" \u002F>\n\n### 更深入的理解\n+ [通过隐式微分优化数百万个超参数](https:\u002F\u002Farxiv.org\u002Fabs\u002F1911.02590) (Jonathan Lorraine 等，AISTATS 2020) [:octocat:](https:\u002F\u002Fgithub.com\u002FMaximeVandegar\u002FPapers-in-100-Lines-of-Code\u002Ftree\u002Fmain\u002FOptimizing_Millions_of_Hyperparameters_by_Implicit_Differentiation) [:book:](.\u002Fcitations\u002Florraine2020optimizing.txt) \n+ [关于过参数化双层优化中的隐式偏差](https:\u002F\u002Fproceedings.mlr.press\u002Fv162\u002Fvicol22a.html) (Paul Vicol 等，ICML 2022) [:book:](.\u002Fcitations\u002Fvicol2022implicit.txt)\n+ [关于蒸馏集合的大小和近似误差](https:\u002F\u002Farxiv.org\u002Fabs\u002F2305.14113) (Alaa Maalouf & Murad Tukan 等，NeurIPS 2023) [:book:](.\u002Fcitations\u002Fmaalouf2023size.txt)\n+ [数据蒸馏的理论研究](https:\u002F\u002Fopenreview.net\u002Fforum?id=dq5QGXGxoJ) (Zachary Izzo 等，NeurIPS 2023研讨会) [:book:](.\u002Fcitations\u002Fizzo2023theo.txt)\n+ [什么是数据蒸馏学习？](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.04284) (William Yang 等，ICML 2024) [:octocat:](https:\u002F\u002Fgithub.com\u002Fprincetonvisualai\u002FWhat-is-Dataset-Distillation-Learning) [:book:](.\u002Fcitations\u002Fyang2024learning.txt)\n+ [缓解数据蒸馏中的偏差](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.06609) (Justin Cui 等，ICML 2024) [:book:](.\u002Fcitations\u002Fcui2024bias.txt)\n+ [从第一原理出发的数据蒸馏：整合核心信息提取与目的性学习](https:\u002F\u002Farxiv.org\u002Fabs\u002F2409.01410) (Vyacheslav Kungurtsev 等，2024年) [:book:](.\u002Fcitations\u002Fkungurtsev2024first.txt)\n+ [信息引导的扩散采样用于数据蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2507.04619) (Linfeng Ye 等，NeurIPS 2025研讨会) [:book:](.\u002Fcitations\u002Fye2025igds.txt)\n+ [基于差异性的数据压缩视角](https:\u002F\u002Farxiv.org\u002Fabs\u002F2509.10367) (Tong Chen 等，2025年) [:book:](.\u002Fcitations\u002Fchen2025discrepancy.txt)\n+ [通过谱滤波理解数据蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2503.01212) (Deyu Bo 等，ICLR 2026) [:book:](.\u002Fcitations\u002Fbo2026unidd.txt)\n+ [针对记忆型数据的数据蒸馏：软标签可能泄露保留教师的知识](https:\u002F\u002Farxiv.org\u002Fabs\u002F2506.14457) (Freya Behrens 等，ICLR 2026) [:book:](.\u002Fcitations\u002Fbehrens2026soft.txt)\n+ [数据蒸馏高效地编码了非线性任务梯度学习中的低维表示](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.14830) (Yuri Kinoshita 等，2026年) [:book:](.\u002Fcitations\u002Fkinoshita2026lowdim.txt)\n\n\u003Ca name=\"label\" \u002F>\n\n### 标签蒸馏\n+ [灵活的数据集蒸馏：学习标签而非图像](https:\u002F\u002Farxiv.org\u002Fabs\u002F2006.08572)（Ondrej Bohdal 等，NeurIPS 2020 工作坊）[:octocat:](https:\u002F\u002Fgithub.com\u002Fondrejbohdal\u002Flabel-distillation) [:book:](.\u002Fcitations\u002Fbohdal2020flexible.txt)\n+ [软标签数据集蒸馏与文本数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F1910.02551)（Ilia Sucholutsky 等，IJCNN 2021）[:octocat:](https:\u002F\u002Fgithub.com\u002Filia10000\u002Fdataset-distillation) [:book:](.\u002Fcitations\u002Fsucholutsky2021soft.txt)\n+ [在数据集蒸馏中，一个标签胜过千张图片](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.10485)（Tian Qin 等，NeurIPS 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Fsunnytqin\u002Fno-distillation) [:book:](.\u002Fcitations\u002Fqin2024label.txt)\n+ [大规模数据集蒸馏是否需要大规模软标签？](https:\u002F\u002Farxiv.org\u002Fabs\u002F2410.15919)（Lingao Xiao 等，NeurIPS 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Fhe-y\u002Fsoft-label-pruning-for-dataset-distillation) [:book:](.\u002Fcitations\u002Fxiao2024soft.txt)\n+ [DRUPI：利用特权信息进行数据集缩减](https:\u002F\u002Farxiv.org\u002Fabs\u002F2410.01611)（Shaobo Wang 等，2024年）[:book:](.\u002Fcitations\u002Fwang2024drupi.txt)\n+ [标签增强型数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2409.16239)（Seoungyoon Kang & Youngsun Lim 等，WACV 2025）[:book:](.\u002Fcitations\u002Fkang2024label.txt)\n+ [GIFT：以近乎零成本释放蒸馏数据集中标签的全部潜力](https:\u002F\u002Farxiv.org\u002Fabs\u002F2405.14736)（Xinyi Shang & Peng Sun 等，ICLR 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002FLINs-lab\u002FGIFT) [:book:](.\u002Fcitations\u002Fshang2025gift.txt)\n+ [重标签出局！通过轻量化标签空间实现数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2408.08201)（Ruonan Yu 等，ICCV 2025）[:book:](.\u002Fcitations\u002Fyu2025helio.txt)\n\n\u003Ca name=\"quant\" \u002F>\n\n### 数据集量化\n+ [数据集量化](https:\u002F\u002Farxiv.org\u002Fabs\u002F2308.10524)（Daquan Zhou & Kai Wang & Jianyang Gu 等，ICCV 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002Fmagic-research\u002FDataset_Quantization) [:book:](.\u002Fcitations\u002Fzhou2023dataset.txt)\n+ [基于主动学习的自适应采样的数据集量化](https:\u002F\u002Farxiv.org\u002Fabs\u002F2407.07268)（Zhenghao Zhao 等，ECCV 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Fichbill\u002FDQAS) [:book:](.\u002Fcitations\u002Fzhao2024dqas.txt)\n+ [自适应数据集量化](https:\u002F\u002Fwww.arxiv.org\u002Fabs\u002F2412.16895)（Muquan Li 等，AAAI 2025）[:book:](.\u002Fcitations\u002Fli2025adq.txt)\n+ [数据集颜色量化：面向训练的数据集级压缩框架](https:\u002F\u002Farxiv.org\u002Fabs\u002F2602.20650)（Chenyue Yu 等，ICLR 2026）[:book:](.\u002Fcitations\u002Fyu2026dcq.txt)\n\n\u003Ca name=\"decouple\" \u002F>\n\n### 解耦蒸馏\n+ [挤压、恢复与重标签：从全新视角实现 ImageNet 规模的数据集凝缩](https:\u002F\u002Farxiv.org\u002Fabs\u002F2306.13092)（Zeyuan Yin & Zhiqiang Shen 等，NeurIPS 2023）[:globe_with_meridians:](https:\u002F\u002Fzeyuanyin.github.io\u002Fprojects\u002FSRe2L\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002FVILA-Lab\u002FSRe2L\u002Ftree\u002Fmain\u002FSRe2L) [:book:](.\u002Fcitations\u002Fyin2023sre2l.txt)\n+ [大数据时代下的课程式数据合成数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2311.18838)（Zeyuan Yin 等，TMLR 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002FVILA-Lab\u002FSRe2L\u002Ftree\u002Fmain\u002FCDA) [:book:](.\u002Fcitations\u002Fyin2024cda.txt)\n+ [通过多种骨干网络和统计匹配实现的大规模数据凝缩](https:\u002F\u002Farxiv.org\u002Fabs\u002F2311.17950)（Shitong Shao 等，CVPR 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Fshaoshitong\u002FG_VBSM_Dataset_Condensation) [:book:](.\u002Fcitations\u002Fshao2024gvbsm.txt)\n+ [关于蒸馏数据集的多样性和真实性：一种高效的数据集蒸馏范式](https:\u002F\u002Farxiv.org\u002Fabs\u002F2312.03526)（Peng Sun 等，CVPR 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002FLINs-lab\u002FRDED) [:book:](.\u002Fcitations\u002Fsun2024rded.txt)\n+ [信息补偿：任何规模数据集蒸馏的修复方案](https:\u002F\u002Fopenreview.net\u002Fforum?id=2SnmKd1JK4)（Peng Sun 等，ICLR 2024研讨会）[:book:](.\u002Fcitations\u002Fsun2024lic.txt)\n+ [阐明数据集凝缩的设计空间](https:\u002F\u002Farxiv.org\u002Fabs\u002F2404.13733)（Shitong Shao 等，NeurIPS 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Fshaoshitong\u002FEDC) [:book:](.\u002Fcitations\u002Fshao2024edc.txt)\n+ [多样性驱动的合成：通过定向权重调整提升数据集蒸馏效果](https:\u002F\u002Farxiv.org\u002Fabs\u002F2409.17612)（Jiawei Du 等，NeurIPS 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002FAngusDujw\u002FDiversity-Driven-Synthesis) [:book:](.\u002Fcitations\u002Fdu2024diversity.txt)\n+ [打破类别壁垒：通过跨类特征补偿器实现高效数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2408.06927)（Xin Zhang 等，ICLR 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002Fzhangxin-xd\u002FUFC) [:book:](.\u002Fcitations\u002Fzhang2025infer.txt)\n+ [DELT：一种简单的多样性驱动的早晚期训练用于数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2411.19946)（Zhiqiang Shen & Ammar Sherif 等，CVPR 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002FVILA-Lab\u002FDELT) [:book:](.\u002Fcitations\u002Fshen2025delt.txt)\n+ [通过非关键区域精炼提升数据集蒸馏效果](https:\u002F\u002Farxiv.org\u002Fabs\u002F2503.18267)（Minh-Tuan Tran 等，CVPR 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002Ftmtuan1307\u002FNRR-DD) [:book:](.\u002Fcitations\u002Ftran2025nrrdd.txt)\n+ [课程式数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2405.09150)（Zhiheng Ma & Anjia Cao 等，TIP 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002FMIV-XJTU\u002FCUDD) [:book:](.\u002Fcitations\u002Fma2025cudd.txt)\n+ [FADRM：用于数据集蒸馏的快速准确的数据残差匹配](https:\u002F\u002Farxiv.org\u002Fabs\u002F2506.24125)（Jiacheng Cui & Xinyue Bi 等，NeurIPS 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002FJiacheng8\u002FFADRM) [:book:](.\u002Fcitations\u002Fcui2025fadrm.txt)\n+ [FocusDD：注入真实场景以实现稳健的数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2501.06405)（Youbin Hu 等，2025年）[:book:](.\u002Fcitations\u002Fhu2025focusdd.txt)\n+ [通过委员会投票进行数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2501.07575)（Jiacheng Cui 等，2025年）[:octocat:](https:\u002F\u002Fgithub.com\u002FJiacheng8\u002FCV-DD) [:book:](.\u002Fcitations\u002Fcui2025cvdd.txt)\n+ [PRISM：通过解耦架构先验来多样化数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.09905)（Brian B. Moser 等，TMLR 2026）[:octocat:](https:\u002F\u002Fgithub.com\u002FBrian-Moser\u002Fprism) [:book:](.\u002Fcitations\u002Fmoser2026prism.txt)\n+ [DiRe：促进多样性的数据集凝缩正则化](https:\u002F\u002Farxiv.org\u002Fabs\u002F2512.13083)（Saumyaranjan Mohanty 等，WACV 2026）[:octocat:](https:\u002F\u002Fgithub.com\u002FDIL-IITH\u002FDiRe) [:book:](.\u002Fcitations\u002Fmohanty2026dire.txt)\n+ [在数据集蒸馏中夯实并提升信息量与实用性](https:\u002F\u002Farxiv.org\u002Fabs\u002F2601.21296)（Shaobo Wang 等，ICLR 2026）[:book:](.\u002Fcitations\u002Fwang2026infoutil.txt)\n+ [固定锚点还不够：动态检索与持久同调用于数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2602.24144)（Muquan Li 等，CVPR 2026）[:book:](.\u002Fcitations\u002Fli2026reta.txt)\n\n\u003Ca name=\"multi\" \u002F>\n\n### 多模态蒸馏\n+ [视觉-语言数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2308.07545)（吴欣迪等，TMLR 2024）[:globe_with_meridians:](https:\u002F\u002Fprincetonvisualai.github.io\u002Fmultimodal_dataset_distillation\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fprincetonvisualai\u002Fmultimodal_dataset_distillation) [:book:](.\u002Fcitations\u002Fwu2024multi.txt)\n+ [基于低秩相似性挖掘的多模态数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.03793)（许悦等，ICML 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Fsilicx\u002FLoRS_Distill) [:book:](.\u002Fcitations\u002Fxu2024lors.txt)\n+ [视听数据集蒸馏](https:\u002F\u002Fopenreview.net\u002Fforum?id=IJlbuSrXmk)（萨克沙姆·辛格·库什瓦哈等，TMLR 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Fsakshamsingh1\u002FAVDD) [:book:](.\u002Fcitations\u002Fkush2024avdd.txt)\n+ [超越模态坍塌：用于多模态数据集蒸馏的表征融合](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.14705)（张鑫等，NeurIPS 2025）[:book:](.\u002Fcitations\u002Fzhang2025mdd.txt)\n+ [基于生成模型的高效多模态数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2509.15472)（赵正浩等，NeurIPS 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002Fichbill\u002FEDGE) [:book:](.\u002Fcitations\u002Fzhao2025edge.txt)\n+ [CovMatch：基于交叉协方差引导、可训练文本编码器的多模态数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2510.18583)（李勇民等，NeurIPS 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002FYongalls\u002FCovMatch) [:book:](.\u002Fcitations\u002Flee2025covmatch.txt)\n+ [解耦的视听数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.17890)（李文渊、李广等，2025年）[:book:](.\u002Fcitations\u002Fli2025davdd.txt)\n+ [ImageBindDC：基于ImageBind的压缩技术对多模态数据进行精简](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.08263)（闵岳、王绍博等，AAAI 2026）[:book:](.\u002Fcitations\u002Fmin2026imagebinddc.txt)\n+ [原型引导的数据合成使多模态数据集蒸馏更简单](https:\u002F\u002Farxiv.org\u002Fabs\u002F2602.19756)（崔俊赫等，ICLR 2026）[:octocat:](https:\u002F\u002Fgithub.com\u002Fjunhyeok9712\u002FPDS) [:book:](.\u002Fcitations\u002Fchoi2026multi.txt)\n+ [分阶段教师模型驱动的多模态数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.25388)（郭圣斌、赵航等，ICLR 2026）[:octocat:](https:\u002F\u002Fgithub.com\u002FPrevisior\u002FPTM-ST) [:book:](.\u002Fcitations\u002Fguo2026ptmst.txt)\n+ [多模态数据集蒸馏中的异步匹配与动态采样](https:\u002F\u002Fopenreview.net\u002Fforum?id=7SgSMKM2KF)（齐丁等，ICLR 2026）[:book:](.\u002Fcitations\u002Fqi2026amd.txt)\n\n\u003Ca name=\"self\" \u002F>\n\n### 自监督蒸馏\n+ [用于迁移学习的自监督数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.06511)（李东朴、李西妮等，ICLR 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Fdb-Lee\u002Fselfsup_dd) [:book:](.\u002Fcitations\u002Flee2024self.txt)\n+ [免费的效率：理想的数据就是可迁移的表征](https:\u002F\u002Farxiv.org\u002Fabs\u002F2405.14669)（孙鹏等，NeurIPS 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002FLINs-lab\u002FReLA) [:book:](.\u002Fcitations\u002Fsun2024rela.txt)\n+ [自监督数据集蒸馏：只需良好的压缩即可](https:\u002F\u002Farxiv.org\u002Fabs\u002F2404.07976)（周木欣等，2024年）[:octocat:](https:\u002F\u002Fgithub.com\u002FVILA-Lab\u002FSRe2L\u002Ftree\u002Fmain\u002FSCDD\u002F) [:book:](.\u002Fcitations\u002Fzhou2024self.txt)\n+ [通过知识蒸馏实现数据集蒸馏：迈向深度网络的高效自监督预训练](https:\u002F\u002Farxiv.org\u002Fabs\u002F2410.02116)（乔希·西达尔特等，ICLR 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002Fjiayini1119\u002FMKDT) [:book:](.\u002Fcitations\u002Fjoshi2025kd.txt)\n+ [通过参数化、预定义增强和近似方法提升自监督数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2507.21455)（余盛峰等，ICLR 2025）[:book:](.\u002Fcitations\u002Fyu2025self.txt)\n+ [面向预训练自监督视觉模型的数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.16674)（乔治·卡泽纳韦特等，NeurIPS 2025）[:globe_with_meridians:](https:\u002F\u002Flinear-gradient-matching.github.io\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002FGeorgeCazenavette\u002Flinear-gradient-matching) [:book:](.\u002Fcitations\u002Fcazenavette2025dataset.txt)\n\n\u003Ca name=\"benchmark\" \u002F>\n\n### 基准测试\n+ [DC-BENCH：数据精简基准测试](https:\u002F\u002Farxiv.org\u002Fabs\u002F2207.09639)（贾斯汀·崔等，NeurIPS 2022）[:globe_with_meridians:](https:\u002F\u002Fdc-bench.github.io\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fjustincui03\u002Fdc_benchmark) [:book:](.\u002Fcitations\u002Fcui2022dc.txt)\n+ [关于数据集蒸馏的全面研究：性能、隐私、鲁棒性和公平性](https:\u002F\u002Farxiv.org\u002Fabs\u002F2305.03355)（陈宗雄、耿嘉慧等，2023年）[:book:](.\u002Fcitations\u002Fchen2023study.txt)\n+ [BEARD：针对数据集蒸馏的对抗鲁棒性基准测试](https:\u002F\u002Farxiv.org\u002Fabs\u002F2411.09265)（周政等，2024年）[:globe_with_meridians:](https:\u002F\u002Fbeard-leaderboard.github.io\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fzhouzhengqd\u002FBEARD\u002F) [:book:](.\u002Fcitations\u002Fzhou2024beard.txt)\n+ [DD-RobustBench：数据集蒸馏的对抗鲁棒性基准测试](https:\u002F\u002Farxiv.org\u002Fabs\u002F2403.13322)（吴一凡等，TIP 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002FFredWU-HUST\u002FDD-RobustBench) [:book:](.\u002Fcitations\u002Fwu2025robust.txt)\n+ [DD-Ranking：重新思考数据集蒸馏的评估](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.13300)（李泽凯、钟新豪等，2025年）[:globe_with_meridians:](https:\u002F\u002Fnus-hpc-ai-lab.github.io\u002FDD-Ranking\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002FNUS-HPC-AI-Lab\u002FDD-Ranking) [:book:](.\u002Fcitations\u002Fli2025ranking.txt)\n+ [修正的解耦数据集蒸馏：为公平且全面的评估提供更深入的视角](https:\u002F\u002Farxiv.org\u002Fabs\u002F2509.19743)（钟新豪等，ICLR 2026）[:book:](.\u002Fcitations\u002Fzhong2026rd3.txt)\n\n\u003Ca name=\"survey\" \u002F>\n\n### 综述\n+ [数据蒸馏：综述](https:\u002F\u002Farxiv.org\u002Fabs\u002F2301.04272)（萨奇德瓦·诺文等，TMLR 2023）[:book:](.\u002Fcitations\u002Fsachdeva2023survey.txt)\n+ [关于数据集蒸馏的综述：方法、应用及未来方向](https:\u002F\u002Farxiv.org\u002Fabs\u002F2305.01975)（耿嘉慧、陈宗雄等，IJCAI 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002FGuang000\u002FAwesome-Dataset-Distillation) [:book:](.\u002Fcitations\u002Fgeng2023survey.txt)\n+ [关于数据集蒸馏的全面综述](https:\u002F\u002Farxiv.org\u002Fabs\u002F2301.05603)（雷世业等，TPAMI 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002FGuang000\u002FAwesome-Dataset-Distillation) [:book:](.\u002Fcitations\u002Flei2023survey.txt)\n+ [数据集蒸馏：全面回顾](https:\u002F\u002Farxiv.org\u002Fabs\u002F2301.07014)（于若楠、刘松华等，TPAMI 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002FGuang000\u002FAwesome-Dataset-Distillation) [:book:](.\u002Fcitations\u002Fyu2023review.txt)\n+ [数据集蒸馏的发展历程：迈向可扩展且通用的解决方案](https:\u002F\u002Farxiv.org\u002Fabs\u002F2502.05673)（刘平等，2025年）[:book:](.\u002Fcitations\u002Fliu2025survey.txt)\n\n\u003Ca name=\"thesis\" \u002F>\n\n### 博士论文\n+ [利用数据精简实现神经网络的高效训练](https:\u002F\u002Fera.ed.ac.uk\u002Fhandle\u002F1842\u002F39756)（赵波，爱丁堡大学，2023年）[:book:](.\u002Fcitations\u002Fzhao2023thesis.txt)\n\n\u003Ca name=\"workshop\" \u002F>\n\n### 研讨会\n+ 第1届CVPR数据集蒸馏研讨会（Saeed Vahidian等，CVPR 2024）[:globe_with_meridians:](https:\u002F\u002Fsites.google.com\u002Fview\u002Fdd-cvpr2024\u002Fhome)\n\n\u003Ca name=\"challenge\" \u002F>\n\n### 挑战赛\n+ 第一届数据集蒸馏挑战赛（Kai Wang & Ahmad Sajedi等，ECCV 2024）[:globe_with_meridians:](https:\u002F\u002Fwww.dd-challenge.com\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002FDataDistillation\u002FECCV2024-Dataset-Distillation-Challenge)\n\n## 应用\n\n\u003Ca name=\"continual\" \u002F>\n\n### 持续学习\n+ [利用合成数据学习减少灾难性遗忘]（Wojciech Masarczyk等，CVPR 2020研讨会）[:book:](.\u002Fcitations\u002Fmasarczyk2020reducing.txt)\n+ [凝聚型复合记忆持续学习]（Felix Wiewel等，IJCNN 2021）[:octocat:](https:\u002F\u002Fgithub.com\u002FFelixWiewel\u002FCCMCL) [:book:](.\u002Fcitations\u002Fwiewel2021soft.txt)\n+ [蒸馏回放：通过合成样本克服遗忘]（Andrea Rosasco等，IJCAI 2021研讨会）[:octocat:](https:\u002F\u002Fgithub.com\u002Fandrearosasco\u002FDistilledReplay) [:book:](.\u002Fcitations\u002Frosasco2021distilled.txt)\n+ [在线持续学习中的样本凝聚]（Mattia Sangermano等，IJCNN 2022）[:octocat:](https:\u002F\u002Fgithub.com\u002FMattiaSangermano\u002FOLCGM) [:book:](.\u002Fcitations\u002Fsangermano2022sample.txt)\n+ [高效数据集凝聚插件及其在持续学习中的应用]（Enneng Yang等，NeurIPS 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002FEnnengYang\u002FAn-Efficient-Dataset-Condensation-Plugin) [:book:](.\u002Fcitations\u002Fyang2023efficient.txt)\n+ [面向内存受限的在线持续学习的流式数据摘要]（Jianyang Gu等，AAAI 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Fvimar-gu\u002FSSD) [:book:](.\u002Fcitations\u002Fgu2024ssd.txt)\n+ [CD2：用于少样本类增量学习的约束数据集蒸馏]（Kexin Bao等，IJCAI 2025）[:book:](.\u002Fcitations\u002Fbao2025cd2.txt)\n+ [面向领域增量数据集蒸馏的非对称合成数据更新]（Minyoung Oh等，ICLR 2026）[:octocat:](https:\u002F\u002Fgithub.com\u002Fmyoh97\u002FDIDD-ASU) [:book:](.\u002Fcitations\u002Foh2026asu.txt)\n\n\u003Ca name=\"privacy\" \u002F>\n\n### 隐私\n+ [免费的隐私：数据集凝聚如何助力隐私保护？]（Tian Dong等，ICML 2022）[:book:](.\u002Fcitations\u002Fdong2022privacy.txt)\n+ [带有判别信息的隐私集合生成]（Dingfan Chen等，NeurIPS 2022）[:octocat:](https:\u002F\u002Fgithub.com\u002FDingfanChen\u002FPrivate-Set) [:book:](.\u002Fcitations\u002Fchen2022privacy.txt)\n+ [“免费的隐私”并非无代价：数据集凝聚如何助力隐私保护？]（Nicholas Carlini等，2022年）[:book:](.\u002Fcitations\u002Fcarlini2022no.txt)\n+ [针对数据集蒸馏的后门攻击]（Yugeng Liu等，NDSS 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002Fliuyugeng\u002Fbaadd) [:book:](.\u002Fcitations\u002Fliu2023backdoor.txt)\n+ [差分隐私核诱导点（DP-KIP）用于隐私保护的数据蒸馏]（Margarita Vinaroz等，2023年）[:octocat:](https:\u002F\u002Fgithub.com\u002Fdpclip\u002Fdpclip) [:book:](.\u002Fcitations\u002Fvinaroz2023dpkip.txt)\n+ [利用神经切空间核与数据集蒸馏理解重构攻击]（Noel Loo等，ICLR 2024）[:book:](.\u002Fcitations\u002Floo2024attack.txt)\n+ [重新思考数据集蒸馏上的后门攻击：基于核方法的视角]（Ming-Yu Chung等，ICLR 2024）[:book:](.\u002Fcitations\u002Fchung2024backdoor.txt)\n+ [差分隐私数据集凝聚]（Zheng等，NDSS 2024研讨会）[:book:](.\u002Fcitations\u002Fzheng2024differentially.txt)\n+ [面向联邦学习的数据集蒸馏自适应后门攻击]（Ze Chai等，ICC 2024）[:book:](.\u002Fcitations\u002Fchai2024backdoor.txt)\n+ [提升隐私保护数据集蒸馏中的噪声效率]（Runkai Zheng等，ICCV 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002Fhumansensinglab\u002FDosser) [:book:](.\u002Fcitations\u002Fzheng2025dosser.txt)\n+ [SNEAKDOOR：针对基于分布匹配的数据集凝聚的隐蔽后门攻击]（He Yang & Dongyi Lv等，NeurIPS 2025）[:book:](.\u002Fcitations\u002Fyang2025sneakdoor.txt)\n+ [毒化蒸馏：无需原始数据访问即可向蒸馏数据集中注入后门]（Ziyuan Yang等，AAAI 2026）[:book:](.\u002Fcitations\u002Fyang2026pd.txt)\n+ [DP-GENG：由差分隐私生成的数据引导的差分隐私数据集蒸馏]（Shuo Shi等，AAAI 2026）[:book:](.\u002Fcitations\u002Fshi2026dpgeng.txt）\n\n\u003Ca name=\"medical\" \u002F>\n\n### 医疗\n+ [软标签匿名胃部X光图像蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2104.02857)（Guang Li等，ICIP 2020）[:octocat:](https:\u002F\u002Fgithub.com\u002FGuang000\u002Fdataset-distillation) [:book:](.\u002Fcitations\u002Fli2020soft.txt) \n+ [基于软标签数据集蒸馏的压缩胃部图像生成用于医疗数据共享](https:\u002F\u002Farxiv.org\u002Fabs\u002F2209.14635)（Guang Li等，CMPB 2022）[:octocat:](https:\u002F\u002Fgithub.com\u002FGuang000\u002Fdataset-distillation) [:book:](.\u002Fcitations\u002Fli2022compressed.txt)\n+ [用于医疗数据集共享的数据集蒸馏](https:\u002F\u002Fr2hcai.github.io\u002FAAAI-23\u002Fpages\u002Faccepted-papers.html)（Guang Li等，AAAI 2023研讨会）[:octocat:](https:\u002F\u002Fgithub.com\u002FGuang000\u002Fmtt-distillation) [:book:](.\u002Fcitations\u002Fli2023sharing.txt)\n+ [具有可泛化数据集蒸馏的通信高效联邦皮肤病变分类](https:\u002F\u002Flink.springer.com\u002Fchapter\u002F10.1007\u002F978-3-031-47401-9_2)（Yuchen Tian & Jiacheng Wang等，MICCAI 2023研讨会）[:book:](.\u002Fcitations\u002Ftian2023gdd.txt)\n+ [重要性感知自适应数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2401.15863)（Guang Li等，NN 2024）[:book:](.\u002Fcitations\u002Fli2024iadd.txt)\n+ [用于组织病理学安全数据共享的图像蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.13536)（Zhe Li等，MICCAI 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002FZheLi2020\u002FInfoDist) [:book:](.\u002Fcitations\u002Fli2024infodist.txt)\n+ [MedSynth：利用生成模型进行医疗数据共享](https:\u002F\u002Flink.springer.com\u002Fchapter\u002F10.1007\u002F978-3-031-72390-2_61)（Renuga Kanagavelu等，MICCAI 2024）[:book:](.\u002Fcitations\u002Fkanagavelu2024medsynth.txt)\n+ [用于医疗数据集蒸馏的渐进式轨迹匹配](https:\u002F\u002Farxiv.org\u002Fabs\u002F2403.13469)（Zhen Yu等，2024年）[:book:](.\u002Fcitations\u002Fyu2024progressive.txt)\n+ [医学影像中的数据集蒸馏：可行性研究](https:\u002F\u002Farxiv.org\u002Fabs\u002F2407.14429)（Muyang Li等，2024年）[:book:](.\u002Fcitations\u002Fli2024medical.txt)\n+ [用于组织病理学图像分类的数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2408.09709)（Cong Cong等，2024年）[:book:](.\u002Fcitations\u002Fcong2024dataset.txt)\n+ [面向医学图像分析的多模态视觉预训练](https:\u002F\u002Farxiv.org\u002Fabs\u002F2410.10604)（Shaohao Rui & Lingzhi Chen等，CVPR 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002Fopenmedlab\u002FBrainMVP) [:book:](.\u002Fcitations\u002Frui2025brain.txt)\n+ [FedWSIDD：通过数据集蒸馏实现的联邦全切片图像分类](https:\u002F\u002Farxiv.org\u002Fabs\u002F2506.15365)（Haolong Jin等，MICCAI 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002Ff1oNae\u002FFedWSIDD) [:book:](.\u002Fcitations\u002Fjin2025fedwsidd.txt)\n+ [用于医学图像数据集蒸馏的高阶渐进式轨迹匹配](https:\u002F\u002Farxiv.org\u002Fabs\u002F2509.24177)（Le Dong等，MICCAI 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002FBian-jh\u002FHoP-TM) [:book:](.\u002Fcitations\u002Fdong2025hop.txt)\n+ [用于医学图像增强的低层次数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.13106)（Fengzhi Xu等，2025年）[:book:](.\u002Fcitations\u002Fxu2025low.txt)\n\n\u003Ca name=\"fed\" \u002F>\n\n### 联邦学习\n+ [通过合成数据的联邦学习](https:\u002F\u002Farxiv.org\u002Fabs\u002F2008.04489)（Jack Goetz等，2020年）[:book:](.\u002Fcitations\u002Fgoetz2020federated.txt)\n+ [蒸馏式一次性联邦学习](https:\u002F\u002Farxiv.org\u002Fabs\u002F2009.07999)（Yanlin Zhou等，2020年）[:book:](.\u002Fcitations\u002Fzhou2020distilled.txt)\n+ [DENSE：无数据的一次性联邦学习](https:\u002F\u002Farxiv.org\u002Fabs\u002F2112.12371)（Jie Zhang & Chen Chen等，NeurIPS 2022）[:octocat:](https:\u002F\u002Fgithub.com\u002Fzj-jayzhang\u002FDENSE) [:book:](.\u002Fcitations\u002Fzhang2022dense.txt)\n+ [FedSynth：联邦学习中通过合成数据进行梯度压缩](https:\u002F\u002Farxiv.org\u002Fabs\u002F2204.01273)（Shengyuan Hu等，2022年）[:book:](.\u002Fcitations\u002Fhu2022fedsynth.txt)\n+ [面向联邦学习的元知识凝结](https:\u002F\u002Farxiv.org\u002Fabs\u002F2209.14851)（Ping Liu等，ICLR 2023）[:book:](.\u002Fcitations\u002Fliu2023meta.txt)\n+ [DYNAFED：用全局动态应对客户端数据异质性](https:\u002F\u002Farxiv.org\u002Fabs\u002F2211.10878)（Renjie Pi等，CVPR 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002Fpipilurj\u002Fdynafed) [:book:](.\u002Fcitations\u002Fpi2023dynafed.txt)\n+ [FedDM：用于通信高效的联邦学习的迭代分布匹配](https:\u002F\u002Farxiv.org\u002Fabs\u002F2207.09653)（Yuanhao Xiong & Ruochen Wang等，CVPR 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002Fanonymifish\u002Ffed-distribution-matching) [:book:](.\u002Fcitations\u002Fxiong2023feddm.txt)\n+ [在资源受限的边缘环境中通过去中心化数据集蒸馏进行联邦学习](https:\u002F\u002Farxiv.org\u002Fabs\u002F2208.11311)（Rui Song等，IJCNN 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002Frruisong\u002Ffedd3) [:book:](.\u002Fcitations\u002Fsong2023federated.txt)\n+ [FedLAP-DP：通过分享差分隐私损失近似值进行联邦学习](https:\u002F\u002Farxiv.org\u002Fabs\u002F2302.01068)（Hui-Po Wang等，2023年）[:octocat:](https:\u002F\u002Fgithub.com\u002Fa514514772\u002Ffedlap-dp) [:book:](.\u002Fcitations\u002Fwang2023fed.txt)\n+ [在虚拟异质数据上使用本地-全局蒸馏的联邦学习](https:\u002F\u002Farxiv.org\u002Fabs\u002F2303.02278)（Chun-Yin Huang等，TMLR 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Fubc-tea\u002FFedLGD) [:book:](.\u002Fcitations\u002Fhuang2024federated.txt)\n+ [一种无需聚合的联邦学习，用于应对数据异质性](https:\u002F\u002Farxiv.org\u002Fabs\u002F2404.18962)（Yuan Wang等，CVPR 2024）[:book:](.\u002Fcitations\u002Fwang2024fed.txt)\n+ [通过合成锚点克服去中心化联邦学习中的数据和模型异质性](https:\u002F\u002Farxiv.org\u002Fabs\u002F2405.11525)（Chun-Yin Huang等，ICML 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Fubc-tea\u002FDESA) [:book:](.\u002Fcitations\u002Fhuang2024desa.txt)\n+ [DCFL：非IID感知数据凝结辅助的联邦学习](https:\u002F\u002Fieeexplore.ieee.org\u002Fdocument\u002F10650791)（Xingwang Wang等，IJCNN 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002FJLUssh\u002FDCFL) [:book:](.\u002Fcitations\u002Fwang2024dcfl.txt)\n+ [释放联邦学习潜力：通过深度生成潜变量进行数据集蒸馏的交响曲](https:\u002F\u002Farxiv.org\u002Fabs\u002F2312.01537)（Yuqi Jia & Saeed Vahidian等，ECCV 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002FFedDG23\u002FFedDG-main) [:book:](.\u002Fcitations\u002Fjia2024feddg.txt)\n+ [一次性协作式数据蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2408.02266)（William Holland等，ECAI 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Frayneholland\u002FCollabDM) [:book:](.\u002Fcitations\u002Fholland2024one.txt)\n+ [FedVCK：鲁棒且通信高效的非IID联邦学习，通过有价值的凝结知识用于医学图像分析](https:\u002F\u002Farxiv.org\u002Fabs\u002F2412.18557)（Guochen Yan等，AAAI 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002FYouth-49\u002FFedVCK_2024) [:book:](.\u002Fcitations\u002Fyan2025fedvck.txt)\n\n\n\u003Ca name=\"gnn\" \u002F>\n\n### 图神经网络\n+ [用于图神经网络的图凝聚](https:\u002F\u002Farxiv.org\u002Fabs\u002F2110.07580)（Wei Jin 等，ICLR 2022）[:octocat:](https:\u002F\u002Fgithub.com\u002Fchandlerbang\u002Fgcond) [:book:](.\u002Fcitations\u002Fjin2022graph.txt)\n+ [通过单步梯度匹配凝聚图](https:\u002F\u002Farxiv.org\u002Fabs\u002F2206.07746)（Wei Jin 等，KDD 2022）[:octocat:](https:\u002F\u002Fgithub.com\u002Famazon-research\u002FDosCond) [:book:](.\u002Fcitations\u002Fjin2022condensing.txt)\n+ [通过感受野分布匹配进行图凝聚](https:\u002F\u002Farxiv.org\u002Fabs\u002F2206.13697)（Mengyang Liu 等，2022）[:book:](.\u002Fcitations\u002Fliu2022graph.txt)\n+ [基于核岭回归的图数据集蒸馏](https:\u002F\u002Fdl.acm.org\u002Fdoi\u002F10.1145\u002F3580305.3599398)（Zhe Xu 等，KDD 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002Fpricexu\u002FKIDD) [:book:](.\u002Fcitations\u002Fxu2023kidd.txt)\n+ [无结构图凝聚：从大规模图到凝聚后的无图数据](https:\u002F\u002Farxiv.org\u002Fabs\u002F2306.02664)（Xin Zheng 等，NeurIPS 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002Famanda-zheng\u002Fsfgc) [:book:](.\u002Fcitations\u002Fzheng2023sfgc.txt)\n+ [图数据蒸馏是否与视觉数据集蒸馏类似？](https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.09192)（Beining Yang & Kai Wang 等，NeurIPS 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002FRingBDStack\u002FSGDD) [:book:](.\u002Fcitations\u002Fyang2023sgdd.txt)\n+ [CaT：基于图凝聚的平衡持续图学习](https:\u002F\u002Farxiv.org\u002Fabs\u002F2309.09455)（Yilun Liu 等，ICDM 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002Fsuperallen13\u002FCaT-CGL) [:book:](.\u002Fcitations\u002Fliu2023cat.txt)\n+ [Mirage：面向图分类的模型无关图蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.09486)（Mridul Gupta & Sahil Manchanda 等，ICLR 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Ffrigategnn\u002FMirage) [:book:](.\u002Fcitations\u002Fgupta2024mirage.txt)\n+ [通过特征基匹配进行图蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.09202)（Yang Liu & Deyu Bo 等，ICML 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Fliuyang-tian\u002FGDEM) [:book:](.\u002Fcitations\u002Fliu2024gdem.txt)\n+ [驾驭复杂性：通过扩展窗口匹配实现无损图凝聚](https:\u002F\u002Farxiv.org\u002Fabs\u002F2402.05011)（Yuchen Zhang & Tianle Zhang & Kai Wang 等，ICML 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Fnus-hpc-ai-lab\u002Fgeom) [:book:](.\u002Fcitations\u002Fzhang2024geom.txt)\n+ [通过自表达图结构重建进行图数据凝聚](https:\u002F\u002Farxiv.org\u002Fabs\u002F2403.07294)（Zhanyu Liu & Chaolv Zeng 等，KDD 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Fzclzcl0223\u002FGCSR) [:book:](.\u002Fcitations\u002Fliu2024gcsr.txt)\n+ [两招不迷糊：通过设计合理梯度匹配凝聚图](https:\u002F\u002Farxiv.org\u002Fabs\u002F2402.04924)（Tianle Zhang & Yuchen Zhang & Kai Wang 等，2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Fnus-hpc-ai-lab\u002Fctrl) [:book:](.\u002Fcitations\u002Fzhang2024ctrl.txt)\n\n#### 综述\n+ [图规约的全面综述：稀疏化、粗化与凝聚](https:\u002F\u002Farxiv.org\u002Fabs\u002F2402.03358)（Mohammad Hashemi 等，IJCAI 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002FEmory-Melody\u002Fawesome-graph-reduction) [:book:](.\u002Fcitations\u002Fhashemi2024awesome.txt)\n+ [图凝聚综述](https:\u002F\u002Farxiv.org\u002Fabs\u002F2402.02000)（Hongjia Xu 等，2024）[:octocat:](https:\u002F\u002Fgithub.com\u002FFrostland12138\u002FAwesome-Graph-Condensation) [:book:](.\u002Fcitations\u002Fxu2024survey.txt)\n+ [图凝聚：综述](https:\u002F\u002Farxiv.org\u002Fabs\u002F2401.11720)（Xinyi Gao 等，TKDE 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002Fxygaog\u002Fgraph-condensation-papers) [:book:](.\u002Fcitations\u002Fgao2025graph.txt)\n\n#### 基准测试\n+ [GC-Bench：一个开放且统一的图凝聚基准测试](https:\u002F\u002Farxiv.org\u002Fabs\u002F2407.00615)（Qingyun Sun & Ziying Chen 等，NeurIPS 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002FRingBDStack\u002FGC-Bench) [:book:](.\u002Fcitations\u002Fsun2024gcbench.txt)\n+ [GCondenser：图凝聚的基准测试](https:\u002F\u002Farxiv.org\u002Fabs\u002F2405.14246)（Yilun Liu 等，2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Fsuperallen13\u002FGCondenser) [:book:](.\u002Fcitations\u002Fliu2024gcondenser.txt)\n+ [GC-Bench：带有新见解的图凝聚基准测试框架](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.16715)（Shengbo Gong & Juntong Ni 等，2024）[:octocat:](https:\u002F\u002Fgithub.com\u002FEmory-Melody\u002FGraphSlim) [:book:](.\u002Fcitations\u002Fgong2024graphslim.txt)\n\n#### 关于图蒸馏的主题将不再更新，因为目前已有关于此主题的足够论文和总结性项目。\n\n\u003Ca name=\"nas\" \u002F>\n\n### 神经架构搜索\n+ [生成式教学网络：通过学习生成合成训练数据加速神经架构搜索](https:\u002F\u002Farxiv.org\u002Fabs\u002F1912.07768)（Felipe Petroski Such 等，ICML 2020）[:octocat:](https:\u002F\u002Fgithub.com\u002Fuber-research\u002FGTN) [:book:](.\u002Fcitations\u002Fsuch2020generative.txt)\n+ [利用梯度匹配和隐式微分学习生成合成训练数据](https:\u002F\u002Farxiv.org\u002Fabs\u002F2203.08559)（Dmitry Medvedev 等，AIST 2021）[:octocat:](https:\u002F\u002Fgithub.com\u002Fdm-medvedev\u002Fefficientdistillation) [:book:](.\u002Fcitations\u002Fmedvedev2021tabular.txt)\n+ [校准数据集蒸馏以加快超参数搜索](https:\u002F\u002Farxiv.org\u002Fabs\u002F2405.17535)（Mucong Ding 等，2024）[:book:](.\u002Fcitations\u002Fding2024hcdc.txt)\n\n\u003Ca name=\"fashion\" \u002F>\n\n### 时尚、艺术与设计\n+ [可穿戴 ImageNet：通过数据集蒸馏合成可平铺纹理](https:\u002F\u002Fopenaccess.thecvf.com\u002Fcontent\u002FCVPR2022W\u002FCVFAD\u002Fhtml\u002FCazenavette_Wearable_ImageNet_Synthesizing_Tileable_Textures_via_Dataset_Distillation_CVPRW_2022_paper.html)（George Cazenavette 等，CVPR 2022 工作坊）[:globe_with_meridians:](https:\u002F\u002Fgeorgecazenavette.github.io\u002Fmtt-distillation\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fgeorgecazenavette\u002Fmtt-distillation) [:book:](.\u002Fcitations\u002Fcazenavette2022textures.txt)\n+ [向设计师学习：通过数据集蒸馏分析时尚搭配](https:\u002F\u002Fieeexplore.ieee.org\u002Fdocument\u002F9897234)（Yulan Chen 等，ICIP 2022）[:book:](.\u002Fcitations\u002Fchen2022fashion.txt)\n+ [具有自适应轨迹匹配的银河数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2311.17967)（Haowen Guan 等，NeurIPS 2023 工作坊）[:octocat:](https:\u002F\u002Fgithub.com\u002FHaowenGuan\u002FGalaxy-Dataset-Distillation) [:book:](.\u002Fcitations\u002Fguan2023galaxy.txt)\n\n\u003Ca name=\"rec\" \u002F>\n\n### 推荐系统\n+ [无限推荐网络：一种以数据为中心的方法](https:\u002F\u002Farxiv.org\u002Fabs\u002F2206.02626)（Noveen Sachdeva 等，NeurIPS 2022）[:octocat:](https:\u002F\u002Fgithub.com\u002Fnoveens\u002Fdistill_cf) [:book:](.\u002Fcitations\u002Fsachdeva2022data.txt)\n+ [CTR 预测中类别型数据蒸馏的梯度匹配](https:\u002F\u002Fdl.acm.org\u002Fdoi\u002F10.1145\u002F3604915.3608769)（Chen Wang 等，RecSys 2023）[:book:](.\u002Fcitations\u002Fwang2023cgm.txt)\n+ [TD3：基于 Tucker 分解的序列推荐数据集蒸馏方法](https:\u002F\u002Farxiv.org\u002Fabs\u002F2502.02854)（Jiaqing Zhang 等，WWW 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002FUSTC-StarTeam\u002FTD3) [:book:](.\u002Fcitations\u002Fzhang2025td3.txt)\n+ [DIET：为推荐系统持续学习数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.24958)（Jiaqing Zhang 等，2026）[:book:](.\u002Fcitations\u002Fzhang2026diet.txt)\n\n\u003Ca name=\"blackbox\" \u002F>\n\n### 黑盒优化\n+ [用于离线无限宽模型优化的双向学习](https:\u002F\u002Farxiv.org\u002Fabs\u002F2209.07507)（Can Chen 等，NeurIPS 2022）[:octocat:](https:\u002F\u002Fgithub.com\u002Fggchen1997\u002Fbdi) [:book:](.\u002Fcitations\u002Fchen2022bidirectional.txt) \n+ [用于离线基于模型的生物序列设计的双向学习](https:\u002F\u002Farxiv.org\u002Fabs\u002F2301.02931)（Can Chen 等，ICML 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002FGGchen1997\u002FBIB-ICML2023-Submission) [:book:](.\u002Fcitations\u002Fchen2023bidirectional.txt)\n\n\u003Ca name=\"robustness\" \u002F>\n\n### 鲁棒性\n+ [仅靠数据就能实现鲁棒性吗？](https:\u002F\u002Farxiv.org\u002Fabs\u002F2207.11727)（Nikolaos Tsilivis 等，ICML 2022 工作坊）[:book:](.\u002Fcitations\u002Ftsilivis2022robust.txt)\n+ [迈向鲁棒的数据集学习](https:\u002F\u002Farxiv.org\u002Fabs\u002F2211.10752)（Yihan Wu 等，2022）[:book:](.\u002Fcitations\u002Fwu2022towards.txt)\n+ [重新思考数据蒸馏：不要忽视校准](https:\u002F\u002Farxiv.org\u002Fabs\u002F2307.12463)（Dongyao Zhu 等，ICCV 2023）[:book:](.\u002Fcitations\u002Fzhu2023calibration.txt)\n+ [迈向可信的数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2307.09165)（Shijie Ma 等，PR 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Fmashijie1028\u002FTrustDD\u002F)  [:book:](.\u002Fcitations\u002Fma2024trustworthy.txt)\n+ [通过曲率正则化实现对抗鲁棒的数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2403.10045)（Eric Xue 等，AAAI 2025）[:globe_with_meridians:](https:\u002F\u002Fyumozi.github.io\u002FGUARD\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fyumozi\u002FGUARD) [:book:](.\u002Fcitations\u002Fxue2025robust.txt)\n+ [具有风险最小化的群体分布鲁棒数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2402.04676)（Saeed Vahidian、Mingyu Wang 和 Jianyang Gu 等，ICLR 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002FMming11\u002FRobustDatasetDistillation) [:book:](.\u002Fcitations\u002Fvahidian2025group.txt)\n+ [ROME 在逆境中锻造：通过信息瓶颈实现鲁棒的蒸馏数据集](https:\u002F\u002Fopenreview.net\u002Fforum?id=agtwOsnLUB)（Zheng Zhou 等，ICML 2025）[:globe_with_meridians:](https:\u002F\u002Fzhouzhengqd.github.io\u002Frome.page\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fzhouzhengqd\u002FROME) [:book:](.\u002Fcitations\u002Fzhou2025rome.txt)\n\n\u003Ca name=\"fairness\" \u002F>\n\n### 公平性\n+ [公平图蒸馏](https:\u002F\u002Fopenreview.net\u002Fforum?id=xW0ayZxPWs)（Qizhang Feng 等，NeurIPS 2023）[:book:](.\u002Fcitations\u002Ffeng2023fair.txt)\n+ [FairDD：公平数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2411.19623)（Qihang Zhou 等，NeurIPS 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002Fzqhang\u002FFairDD) [:book:](.\u002Fcitations\u002Fzhou2025fair.txt)\n\n\u003Ca name=\"text\" \u002F>\n\n### 文本\n+ [文本分类中的数据蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2104.08448)（Yongqi Li 等，2021）[:book:](.\u002Fcitations\u002Fli2021text.txt)\n+ [带有注意力标签的数据集蒸馏用于微调 BERT](https:\u002F\u002Faclanthology.org\u002F2023.acl-short.12\u002F)（Aru Maekawa 等，ACL 2023）[:octocat:](https:\u002F\u002Fgithub.com\u002Farumaekawa\u002Fdataset-distillation-with-attention-labels) [:book:](.\u002Fcitations\u002Fmaekawa2023text.txt)\n+ [DiLM：将数据集蒸馏为语言模型以进行文本级数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2404.00264)（Aru Maekawa 等，NAACL 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Farumaekawa\u002FDiLM) [:book:](.\u002Fcitations\u002Fmaekawa2024dilm.txt)\n+ [通过语言模型嵌入进行文本数据集蒸馏](https:\u002F\u002Faclanthology.org\u002F2024.findings-emnlp.733\u002F)（Yefan Tao 等，EMNLP 2024）[:book:](.\u002Fcitations\u002Ftao2024textual.txt)\n+ [UniDetox：通过数据集蒸馏对大型语言模型进行通用去毒](https:\u002F\u002Farxiv.org\u002Fabs\u002F2504.20500)（Huimin Lu 等，ICLR 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002FEminLU\u002FUniDetox) [:book:](.\u002Fcitations\u002Flu2025llm.txt)\n+ [知识层级引导的生物医学数据集蒸馏用于领域 LLM 训练](https:\u002F\u002Farxiv.org\u002Fabs\u002F2501.15108)（Xunxin Cai、Chengrui Wang 和 Qingqing Long 等，DASFAA 2025）[:book:](.\u002Fcitations\u002Fcai2025llm.txt)\n+ [通过梯度匹配生成合成文本以训练大型语言模型](https:\u002F\u002Farxiv.org\u002Fabs\u002F2502.17607)（Dang Nguyen 和 Zeman Li 等，ICML 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002FBigML-CS-UCLA\u002FGRADMM) [:book:](.\u002Fcitations\u002Fnguyen2025llm.txt)\n+ [CondenseLM：通过奖励匹配驱动的 LLM 进行文本数据集浓缩](https:\u002F\u002Faclanthology.org\u002F2025.emnlp-main.65\u002F)（Cheng Shen 等，EMNLP 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002Fcs6331\u002FCondenseLM\u002F) [:book:](.\u002Fcitations\u002Fshen2025llm.txt)\n\n\u003Ca name=\"video\" \u002F>\n\n### 视频\n+ [与静止图像共舞：基于静态-动态解耦的视频蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2312.00362)（Ziyu Wang 和 Yue Xu 等，CVPR 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Fyuz1wan\u002Fvideo_distillation) [:book:](.\u002Fcitations\u002Fwang2023dancing.txt)\n+ [视频集合蒸馏：信息多样化与时间密集化](https:\u002F\u002Farxiv.org\u002Fabs\u002F2412.00111)（Yinjie Zhao 等，2024）[:book:](.\u002Fcitations\u002Fzhao2024video.txt)\n+ [关于视频动作数据集浓缩的大规模研究](https:\u002F\u002Farxiv.org\u002Fabs\u002F2412.21197)（Yang Chen 等，2024）[:octocat:](https:\u002F\u002Fgithub.com\u002FMCG-NJU\u002FVideo-DC) [:book:](.\u002Fcitations\u002Fchen2024video.txt)\n+ [通过生成网络反演压缩动作分割数据集](https:\u002F\u002Farxiv.org\u002Fabs\u002F2503.14112)（Guodong Ding 等，CVPR 2025）[:book:](.\u002Fcitations\u002Fding2025video.txt)\n+ [潜在视频数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2504.17132)（Ning Li 等，CVPR 2025 工作坊）[:octocat:](https:\u002F\u002Fgithub.com\u002Fliningresearch\u002FLatent_Video_Dataset_Distillation) [:book:](.\u002Fcitations\u002Fli2025latent.txt)\n+ [将视频数据集蒸馏成图像](https:\u002F\u002Farxiv.org\u002Fabs\u002F2512.14621)（Zhenghao Zhao 等，2025）[:book:](.\u002Fcitations\u002Fzhao2025video.txt)\n+ [PRISM：通过渐进式精炼和插入稀疏运动来压缩视频数据集](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.22564)（Jaehyun Choi 等，CVPR 2026）[:book:](.\u002Fcitations\u002Fchoi2026prism.txt)\n\n\u003Ca name=\"tabular\" \u002F>\n\n### 表格数据\n+ [表格数据处理时数据蒸馏方法的新特性](https:\u002F\u002Farxiv.org\u002Fabs\u002F2010.09839)（德米特里·梅德韦杰夫等，AIST 2020）[:octocat:](https:\u002F\u002Fgithub.com\u002Fdm-medvedev\u002Fdataset-distillation) [:book:](.\u002Fcitations\u002Fmedvedev2020tabular.txt)\n\n\u003Ca name=\"retrieval\" \u002F>\n\n### 检索\n+ [迈向高效的深度哈希检索：通过特征嵌入匹配凝练你的数据](https:\u002F\u002Farxiv.org\u002Fabs\u002F2305.18076)（冯涛、张杰等，2023年）[:book:](.\u002Fcitations\u002Ffeng2023hash.txt)\n\n\u003Ca name=\"domain\" \u002F>\n\n### 领域适应\n+ [多源领域适应结合数据字典学习实现数据蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2309.07666)（爱德华多·蒙特苏马等，ICASSP 2024）[:book:](.\u002Fcitations\u002Fmontesuma2024multi.txt)\n\n\u003Ca name=\"super\" \u002F>\n\n### 超分辨率\n+ [GSDD：用于图像超分辨率的生成空间数据蒸馏](https:\u002F\u002Fojs.aaai.org\u002Findex.php\u002FAAAI\u002Farticle\u002Fview\u002F28534)（张海宇等，AAAI 2024）[:book:](.\u002Fcitations\u002Fzhang2024super.txt)\n\n\u003Ca name=\"time\" \u002F>\n\n### 时间序列\n+ [基于双域匹配的时间序列分类数据凝缩](https:\u002F\u002Farxiv.org\u002Fabs\u002F2403.07245)（刘占宇等，KDD 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Fzhyliu00\u002FTimeSeriesCond) [:book:](.\u002Fcitations\u002Fliu2024time.txt)\n+ [CondTSF：用于时间序列预测的一行式数据凝缩插件](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.02131)（丁建荣、刘占宇等，NeurIPS 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002FRafaDD\u002FCondTSF) [:book:](.\u002Fcitations\u002Fding2024time.txt)\n+ [少即是多：通过双重模态匹配实现高效的时间序列数据凝缩](https:\u002F\u002Farxiv.org\u002Fabs\u002F2410.20905)（苗浩等，VLDB 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002Fuestc-liuzq\u002FSTdistillation) [:book:](.\u002Fcitations\u002Fmiao2025timedc.txt)\n+ [DDTime：面向时间序列预测的谱对齐与信息瓶颈数据蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.16715)（李宇奇、丁奎业等，2025年）[:book:](.\u002Fcitations\u002Fli2025time.txt)\n+ [面向时间序列预测的谐波数据蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.03760)（洪承河等，AAAI 2026）[:book:](.\u002Fcitations\u002Fhong2026hdt.txt)\n+ [为高效预测而蒸馏时间序列基础模型](https:\u002F\u002Farxiv.org\u002Fabs\u002F2601.12785)（李宇奇、丁奎业等，ICASSP 2026）[:octocat:](https:\u002F\u002Fgithub.com\u002Fitsnotacie\u002FDistilTS-ICASSP2026) [:book:](.\u002Fcitations\u002Fli2026distilts.txt)\n+ [利用二维压缩实现时空预测的有效数据蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.10410)（权泰亨、崔妍洁等，ICDE 2026）[:octocat:](https:\u002F\u002Fgithub.com\u002Fkbrother\u002FSTemDist) [:book:](.\u002Fcitations\u002Fkwon2026effective.txt)\n+ [ShapeCond：面向时间序列分类的快速形状子指导数据凝缩](https:\u002F\u002Farxiv.org\u002Fabs\u002F2602.09008)（彭思佳等，2026年）[:octocat:](https:\u002F\u002Fgithub.com\u002Flunaaa95\u002FShapeCond) [:book:](.\u002Fcitations\u002Fpeng2026shapecond.txt)\n\n\u003Ca name=\"speech\" \u002F>\n\n### 语音\n+ [用于语音情感识别的数据蒸馏生成模型](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.02963)（法比安·里特-古铁雷斯等，Interspeech 2024）[:book:](.\u002Fcitations\u002Ffabian2024speech.txt)\n\n\u003Ca name=\"unlearning\" \u002F>\n\n### 机器遗忘\n+ [基于反向梯度匹配的蒸馏数据模型](https:\u002F\u002Farxiv.org\u002Fabs\u002F2404.14006)（叶静雯等，CVPR 2024）[:book:](.\u002Fcitations\u002Fye2024datamodel.txt)\n+ [由数据凝缩驱动的机器遗忘](https:\u002F\u002Farxiv.org\u002Fabs\u002F2402.00195)（朱奈德·伊克巴尔·汗，2024年）[:octocat:](https:\u002F\u002Fgithub.com\u002Falgebraicdianuj\u002FDC_U) [:book:](.\u002Fcitations\u002Fkhan2024unlearning.txt)\n\n\u003Ca name=\"rl\" \u002F>\n\n### 强化学习\n+ [行为蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2406.15042)（安德烈·卢普等，ICLR 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002Fflairox\u002Fbehaviour-distillation) [:book:](.\u002Fcitations\u002Flupu2024bd.txt)\n+ [面向离线强化学习的数据蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2407.20299)（乔纳森·莱特、刘元哲等，ICML 2024研讨会）[:globe_with_meridians:](https:\u002F\u002Fdatasetdistillation4rl.github.io\u002F) [:octocat:](https:\u002F\u002Fgithub.com\u002Fggflow123\u002FDDRL) [:book:](.\u002Fcitations\u002Flight2024rl.txt)\n+ [离线行为蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2410.22728)（雷世业等，NeurIPS 2024）[:octocat:](https:\u002F\u002Fgithub.com\u002FLeavesLei\u002FOBD) [:book:](.\u002Fcitations\u002Flei2024obl.txt)\n+ [将强化学习蒸馏进单批次数据集](https:\u002F\u002Farxiv.org\u002Fabs\u002F2508.09283)（康纳·威廉姆等，ECAI 2025）[:book:](.\u002Fcitations\u002Fwilhelm2025rl.txt)\n+ [监督与离线强化学习数据蒸馏的算法保证](https:\u002F\u002Farxiv.org\u002Fabs\u002F2512.00536)（阿扬·古普塔等，ICLR 2026）[:book:](.\u002Fcitations\u002Fgupta2026rl.txt)\n\n\u003Ca name=\"long\" \u002F>\n\n### 长尾分布\n+ [长尾数据集的蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2408.14506)（赵正浩、王浩轩等，CVPR 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002Fichbill\u002FLTDD) [:book:](.\u002Fcitations\u002Fzhao2025long.txt)\n+ [修正长尾数据蒸馏中的软标签纠缠偏差](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.17914)（蒋晨阳等，NeurIPS 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002Fj-cyoung\u002FADSA_DD) [:book:](.\u002Fcitations\u002Fjiang2025long.txt)\n+ [重新思考长尾数据蒸馏：一种具有无偏恢复与重标签功能的统一框架](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.18858)（崔晓等，AAAI 2026）[:book:](.\u002Fcitations\u002Fcui2026long.txt)\n\n\u003Ca name=\"noisy\" \u002F>\n\n### 噪声标签学习\n+ [数据蒸馏器在实际应用中是优秀的标签去噪器](https:\u002F\u002Farxiv.org\u002Fabs\u002F2411.11924)（程乐超等，2024年）[:octocat:](https:\u002F\u002Fgithub.com\u002FKciiiman\u002FDD_LNL) [:book:](.\u002Fcitations\u002Fcheng2024noisy.txt)\n+ [利用监督对比学习实现稳健的数据凝缩](https:\u002F\u002Fopenaccess.thecvf.com\u002Fcontent\u002FICCV2025\u002Fhtml\u002FKim_Robust_Dataset_Condensation_using_Supervised_Contrastive_Learning_ICCV_2025_paper.html)（金熙妍等，ICCV 2025）[:octocat:](https:\u002F\u002Fgithub.com\u002FDISL-Lab\u002FRDC-ICCV2025) [:book:](.\u002Fcitations\u002Fkim2025rdc.txt)\n\n\u003Ca name=\"detection\" \u002F>\n\n### 目标检测\n+ [获取与锻造：面向目标检测的有效数据凝缩](https:\u002F\u002Fopenreview.net\u002Fforum?id=m8MElyzuwp)（齐丁等，NeurIPS 2024）[:book:](.\u002Fcitations\u002Fqi2024dcod.txt)\n+ [OD3：面向目标检测的无优化数据蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2506.01942)（萨尔瓦·K·阿尔哈蒂布、艾哈迈德·埃尔哈格里、邵士通等，ICLR 2026）[:octocat:](https:\u002F\u002Fgithub.com\u002FVILA-Lab\u002FOD3) [:book:](.\u002Fcitations\u002Fkhatib2026od3.txt)\n\n\u003Ca name=\"point\" \u002F>\n\n### 点云\n+ [点云数据集蒸馏](https:\u002F\u002Fopenreview.net\u002Fforum?id=Us8v5tDOFd)（Deyu Bo 等，ICML 2025）[:book:](.\u002Fcitations\u002Fbo2025point.txt)\n+ [基于分布匹配的三维点云数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2503.22154)（Jae-Young Yim & Dongwook Kim 等，NeurIPS 2025）[:book:](.\u002Fcitations\u002Fyim2025point.txt)\n+ [通过可学习形状变形实现的基于参数化的三维点云数据集蒸馏](https:\u002F\u002Fopenreview.net\u002Fforum?id=Qe7dKZOtWM)（Dongwook Kim & Jae-Young Yim 等，ICLR 2026）[:book:](.\u002Fcitations\u002Fkim2026pointmorph.txt)\n\n\u003Ca name=\"uni\" \u002F>\n\n### 通用蒸馏\n\n+ [基于任务驱动扩散的通用数据集蒸馏探索](https:\u002F\u002Fopenaccess.thecvf.com\u002Fcontent\u002FCVPR2025\u002Fhtml\u002FQi_Towards_Universal_Dataset_Distillation_via_Task-Driven_Diffusion_CVPR_2025_paper.html)（Ding Qi 等，CVPR 2025）[:book:](.\u002Fcitations\u002Fqi2025unidd.txt)\n\n\u003Ca name=\"snn\" \u002F>\n\n### 脉冲神经网络\n\n+ [从密集事件中学习：通过事件数据集蒸馏实现快速脉冲神经网络训练](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.12095)（Shuhan Ye 等，2025年）[:book:](.\u002Fcitations\u002Fye2025snn.txt)\n\n\u003Ca name=\"eeg\" \u002F>\n\n### 脑电图\n\n+ [EEG-DLite：用于高效大型脑电模型训练的数据集蒸馏](https:\u002F\u002Farxiv.org\u002Fabs\u002F2512.12210)（Yuting Tang 等，AAAI 2026）[:octocat:](https:\u002F\u002Fgithub.com\u002Ft170815518\u002FEEG-DLite) [:book:](.\u002Fcitations\u002Ftang2026eeg.txt)\n\n\u003Ca name=\"finance\" \u002F>\n\n### 金融\n\n+ [基于分层多源数据集蒸馏的金融领域安全且可解释的欺诈检测](https:\u002F\u002Farxiv.org\u002Fabs\u002F2512.21866)（Yiming Qian 等，ICAIFF 2025）[:book:](.\u002Fcitations\u002Fqian2025finance.txt)\n\n\u003Ca name=\"music\" \u002F>\n\n### 音乐\n\n+ [ConceptCaps：用于音乐模型可解释性的蒸馏概念数据集](https:\u002F\u002Farxiv.org\u002Fabs\u002F2601.14157)（Bruno Sienkiewicz 等，2026年）[:book:](.\u002Fcitations\u002Fsienkiewicz2026music.txt)\n\n\u003Ca name=\"rs\" \u002F>\n\n### 遥感\n\n+ [基于判别原型引导扩散的真实感遥感数据集蒸馏探索](https:\u002F\u002Farxiv.org\u002Fabs\u002F2601.15829)（Yonghao Xu 等，2026年）[:book:](.\u002Fcitations\u002Fxu2026rs.txt)\n\n\u003Ca name=\"dr\" \u002F>\n\n### 去雨\n\n+ [UniRain：基于RAG的数据集蒸馏与多目标重加权优化的统一图像去雨方法](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.03967)（Qianfeng Yang 等，CVPR 2026）[:octocat:](https:\u002F\u002Fgithub.com\u002FQianfengY\u002FUniRain) [:book:](.\u002Fcitations\u002Fyang2026unirain.txt)\n\n\u003Ca name=\"fine\" \u002F>\n\n### 细粒度\n\n+ [FD2：专门用于细粒度数据集蒸馏的框架](https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.25144)（Hongxu Ma & Guang Li 等，2026年）[:book:](.\u002Fcitations\u002Fma2026fd2.txt)\n\n## 媒体报道\n+ [精彩数据集蒸馏的开端](https:\u002F\u002Ftwitter.com\u002FTongzhouWang\u002Fstatus\u002F1560043815204970497?cxt=HHwWgoCz9bPlsaYrAAAA)\n+ [2022年8月最受欢迎的人工智能研究](https:\u002F\u002Fwww.libhunt.com\u002Fposts\u002F874974-d-most-popular-ai-research-aug-2022-ranked-based-on-github-stars)\n+ [一个项目帮你了解数据集蒸馏Dataset Distillation](https:\u002F\u002Fwww.jiqizhixin.com\u002Farticles\u002F2022-10-11-22)\n+ [浓缩就是精华：用大一统视角看待数据集蒸馏](https:\u002F\u002Fmp.weixin.qq.com\u002Fs\u002F__IjS0_FMpu35X9cNhNhPg)\n\n## 星标历史\n[![星标历史图表](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FGuang000_Awesome-Dataset-Distillation_readme_127a773fa9fc.png)](https:\u002F\u002Fstar-history.com\u002F#Guang000\u002FAwesome-Dataset-Distillation&Date)\n\n## 引用“Awesome Dataset Distillation”\n如果您觉得本项目对您的研究有所帮助，请使用以下BibTeX条目。\n```\n@misc{li2022awesome,\n  author={Li, Guang and Zhao, Bo and Wang, Tongzhou},\n  title={Awesome Dataset Distillation},\n  howpublished={\\url{https:\u002F\u002Fgithub.com\u002FGuang000\u002FAwesome-Dataset-Distillation}},\n  year={2022}\n}\n```\n\n## 致谢\n我们衷心感谢以下各位的宝贵建议和贡献：[Nikolaos Tsilivis](https:\u002F\u002Fgithub.com\u002FTsili42)、[Wei Jin](https:\u002F\u002Fgithub.com\u002FChandlerBang)、[Yongchao Zhou](https:\u002F\u002Fgithub.com\u002Fyongchao97)、[Noveen Sachdeva](https:\u002F\u002Fgithub.com\u002Fnoveens)、[Can Chen](https:\u002F\u002Fgithub.com\u002FGGchen1997)、[Guangxiang Zhao](https:\u002F\u002Fgithub.com\u002Fzhaoguangxiang)、[Shiye Lei](https:\u002F\u002Fgithub.com\u002FLeavesLei)、[Xinchao Wang](https:\u002F\u002Fsites.google.com\u002Fsite\u002Fsitexinchaowang\u002F)、[Dmitry Medvedev](https:\u002F\u002Fgithub.com\u002Fdm-medvedev)、[Seungjae Shin](https:\u002F\u002Fgithub.com\u002FSJShin-AI)、[Jiawei Du](https:\u002F\u002Fgithub.com\u002FAngusDujw)、[Yidi Jiang](https:\u002F\u002Fgithub.com\u002FJiang-Yidi)、[Xindi Wu](https:\u002F\u002Fgithub.com\u002FXindiWu)、[Guangyi Liu](https:\u002F\u002Fgithub.com\u002Flgy0404)、[Yilun Liu](https:\u002F\u002Fgithub.com\u002Fsuperallen13)、[Kai Wang](https:\u002F\u002Fgithub.com\u002Fkaiwang960112)、[Yue Xu](https:\u002F\u002Fgithub.com\u002Fsilicx)、[Anjia Cao](https:\u002F\u002Fgithub.com\u002FCAOANJIA)、[Jianyang Gu](https:\u002F\u002Fgithub.com\u002Fvimar-gu)、[Yuanzhen Feng](https:\u002F\u002Fgithub.com\u002Ffengyzpku)、[Peng Sun](https:\u002F\u002Fgithub.com\u002Fsp12138)、[Ahmad Sajedi](https:\u002F\u002Fgithub.com\u002FAhmadSajedii)、[Zhihao Sui](https:\u002F\u002Fgithub.com\u002Fsuizhihao)、[Ziyu Wang](https:\u002F\u002Fgithub.com\u002Fyuz1wan)、[Haoyang Liu](https:\u002F\u002Fgithub.com\u002FLiu-Hy)、[Eduardo Montesuma](https:\u002F\u002Fgithub.com\u002Feddardd)、[Shengbo Gong](https:\u002F\u002Fgithub.com\u002Frockcor)、[Zheng Zhou](https:\u002F\u002Fgithub.com\u002Fzhouzhengqd)、[Zhenghao Zhao](https:\u002F\u002Fgithub.com\u002Fichbill)、[Duo Su](https:\u002F\u002Fgithub.com\u002Fsuduo94)、[Tianhang Zheng](https:\u002F\u002Fgithub.com\u002Ftianzheng4)、[Shijie Ma](https:\u002F\u002Fgithub.com\u002Fmashijie1028)、[Wei Wei](https:\u002F\u002Fgithub.com\u002FWeiWeic6222848)、[Yantai Yang](https:\u002F\u002Fgithub.com\u002FHiter-Q)、[Shaobo Wang](https:\u002F\u002Fgithub.com\u002Fgszfwsb)、[Xinhao Zhong](https:\u002F\u002Fgithub.com\u002Fndhg1213)、[Zhiqiang Shen](https:\u002F\u002Fgithub.com\u002Fszq0214)、[Cong Cong](https:\u002F\u002Fgithub.com\u002Fthomascong121)、[Chun-Yin Huang](https:\u002F\u002Fgithub.com\u002Fchunyinhuang)、[Dai Liu](https:\u002F\u002Fgithub.com\u002FNiaLiu)、[Ruonan Yu](https:\u002F\u002Fgithub.com\u002FLexie-YU)、[William Holland](https:\u002F\u002Fgithub.com\u002Frayneholland)、[Saksham Singh Kushwaha](https:\u002F\u002Fgithub.com\u002Fsakshamsingh1)、[Ping Liu](https:\u002F\u002Fgithub.com\u002Fpinglmlcv)、[Wenliang Zhong](https:\u002F\u002Fgithub.com\u002FZhong0x29a)、[Ning Li](https:\u002F\u002Fgithub.com\u002FNing9319)、[Guochen Yan](https:\u002F\u002Fgithub.com\u002FYouth-49)、[Saumyaranjan Mohanty](https:\u002F\u002Fgithub.com\u002Farareddy)以及[Taehyung Kwon](https:\u002F\u002Fgithub.com\u002Fkbrother)。\n\n“Awesome Dataset Distillation”的[主页](https:\u002F\u002Fguang000.github.io\u002FAwesome-Dataset-Distillation\u002F)由[Longzhen Li](https:\u002F\u002Fgithub.com\u002FLOVELESSG)设计，并由[Mingzhuo Li](https:\u002F\u002Fgithub.com\u002FSumomoTaku)维护。","# Awesome-Dataset-Distillation 快速上手指南\n\n**Awesome-Dataset-Distillation** 并非一个单一的 Python 包，而是一个汇集了数据集蒸馏（Dataset Distillation）领域最新论文、代码实现和资源的精选列表。数据集蒸馏旨在合成一个小规模数据集，使得在该小数据集上训练的模型能在原始大规模数据集上取得高性能。\n\n本指南将帮助你利用该仓库找到合适的工具并快速运行一个基础的蒸馏示例。\n\n## 环境准备\n\n在开始之前，请确保你的开发环境满足以下要求：\n\n*   **操作系统**: Linux (推荐 Ubuntu 20.04+) 或 macOS。Windows 用户建议使用 WSL2。\n*   **Python 版本**: 3.8 - 3.10 (大多数深度学习项目对此范围支持最好)。\n*   **GPU**: 推荐使用 NVIDIA GPU (显存建议 8GB 以上)，因为数据集蒸馏涉及大量的梯度计算。\n*   **前置依赖**:\n    *   `git`: 用于克隆仓库。\n    *   `conda` 或 `venv`: 用于管理虚拟环境。\n    *   `PyTorch`: 核心深度学习框架 (版本需与具体选用的子项目匹配，通常建议 1.9.0+)。\n    *   `torchvision`: 用于加载标准数据集 (如 CIFAR-10, ImageNet)。\n\n> **国内加速建议**:\n> *   克隆仓库时若速度慢，可使用 Gitee 镜像（如有）或配置 Git 代理。\n> *   安装 PyTorch 时，优先使用清华源或中科大源。\n\n## 安装步骤\n\n由于该仓库是资源列表，你需要先克隆仓库，然后选择其中一个具体的算法项目（例如经典的 **Dataset Condensation** 或 **MTT**）进行安装。以下以克隆主仓库并安装一个典型子项目为例：\n\n### 1. 克隆主仓库\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002FGuang000\u002FAwesome-Dataset-Distillation.git\ncd Awesome-Dataset-Distillation\n```\n\n### 2. 选择并安装具体算法\n浏览 `README.md` 中的 [Main](#main) 或 [Gradient\u002FTrajectory Matching Surrogate Objective](#gradient-objective) 部分，选择一个带有 `:octocat: Code` 标记的项目。这里以 **Dataset Condensation with Gradient Matching** 为例：\n\n```bash\n# 克隆具体的算法代码库\ngit clone https:\u002F\u002Fgithub.com\u002FVICO-UoE\u002FDatasetCondensation.git\ncd DatasetCondensation\n\n# 创建虚拟环境 (推荐)\nconda create -n distill python=3.9 -y\nconda activate distill\n\n# 使用国内源安装 PyTorch (根据CUDA版本调整，此处以CPU版为例，生产环境请安装GPU版)\n# 访问 https:\u002F\u002Fmirrors.tuna.tsinghua.edu.cn\u002Fhelp\u002Fanaconda\u002F 获取最新镜像命令\npip install torch torchvision torchaudio --index-url https:\u002F\u002Fdownload.pytorch.org\u002Fwhl\u002Fcu118\n\n# 安装其他依赖\npip install -r requirements.txt\n```\n\n## 基本使用\n\n数据集蒸馏的核心流程通常分为两步：**蒸馏过程**（生成合成数据）和 **评估过程**（在合成数据上训练模型并测试）。\n\n以下是一个基于上述 `DatasetCondensation` 项目的最简使用示例，演示如何在 CIFAR-10 数据集上执行梯度匹配蒸馏。\n\n### 1. 准备数据\n确保已下载 CIFAR-10 数据集，或者让脚本自动下载。通常数据会存放在 `.\u002Fdata` 目录下。\n\n### 2. 运行蒸馏脚本\n执行以下命令开始蒸馏。该命令将尝试从真实的 CIFAR-10 训练集中提炼出少量合成图像（例如每类 10 张）。\n\n```bash\npython main.py \\\n  --dataset CIFAR10 \\\n  --model ConvNet \\\n  --method DM \\\n  --ipc 10 \\\n  --epoch 2000 \\\n  --lr_img 0.1 \\\n  --save_dir .\u002Fcheckpoints\n```\n\n**参数说明：**\n*   `--dataset`: 原始数据集名称 (如 `CIFAR10`, `MNIST`)。\n*   `--method`: 蒸馏算法 (如 `DM` 代表 Dataset Distillation, `DC` 代表 Dataset Condensation)。\n*   `--ipc`: Images Per Class，即每类合成的图片数量（决定压缩率）。\n*   `--epoch`: 蒸馏优化的迭代次数。\n*   `--save_dir`: 保存生成的合成数据集路径。\n\n### 3. 评估效果\n蒸馏完成后，使用生成的合成数据集训练一个新的模型并在测试集上验证准确率：\n\n```bash\npython eval.py \\\n  --dataset CIFAR10 \\\n  --model ConvNet \\\n  --data_path .\u002Fcheckpoints\u002Fdm_cifar10_ipc10.pt \\\n  --epoch 100\n```\n\n运行结束后，终端将输出模型在真实测试集上的准确率，以此衡量蒸馏数据集的质量。\n\n> **提示**: 更多高级用法（如针对 ImageNet 的大规模蒸馏、不同架构的适配）请参考具体子项目的 `README` 文件或 Awesome 列表中对应的论文链接。","某医疗 AI 初创团队需要在受限的隐私合规环境下，利用海量患者影像数据训练轻量级诊断模型，并快速验证新的网络架构。\n\n### 没有 Awesome-Dataset-Distillation 时\n- **文献调研如大海捞针**：团队成员需手动在 arXiv 和各大会议中筛选“数据集蒸馏”相关论文，极易遗漏如梯度匹配（Gradient Matching）等关键早期工作或最新的医疗领域应用案例。\n- **技术选型盲目试错**：由于缺乏对 300+ 篇论文的系统分类，团队难以判断哪种算法适合小样本医疗场景，导致在不适用的通用算法上浪费数周算力进行无效实验。\n- **复现门槛极高**：找不到官方代码链接或标准的 BibTeX 引用，研究人员需花费大量时间逆向工程论文细节，甚至因缺少基准对比而无法评估自身模型效果。\n- **应用场景视野狭窄**：仅关注基础的压缩任务，忽略了该技术在持续学习、隐私保护及神经架构搜索（NAS）中的潜力，错失了优化产品合规性与迭代速度的机会。\n\n### 使用 Awesome-Dataset-Distillation 后\n- **一站式权威索引**：直接获取由领域专家维护的最新论文清单，迅速定位到针对医疗数据集蒸馏的开创性研究及 2026 年最新的细粒度蒸馏框架（如 FD2）。\n- **精准算法匹配**：利用清晰的分类体系，快速锁定适合医疗影像的“形状引导凝聚”或“可学习性引导扩散”算法，将技术验证周期从数周缩短至几天。\n- **开箱即用的资源**：每个条目均附带项目主页、GitHub 代码库及标准引用格式，团队能立即复现 SOTA（最先进）结果，建立可靠的性能基线。\n- **激发创新灵感**：通过浏览隐私保护和推荐系统等跨界应用案例，团队成功将蒸馏技术应用于联邦学习场景，在满足数据不出院的前提下提升了模型泛化能力。\n\nAwesome-Dataset-Distillation 将原本分散碎片化的前沿研究转化为结构化的知识资产，让研发团队能从繁琐的文献工作中解脱，专注于核心算法的创新与落地。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FGuang000_Awesome-Dataset-Distillation_e5c26c8b.jpg","Guang000","Guang Li","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002FGuang000_3d3c4efb.jpg","Assistant Professor at Hokkaido University","Hokkaido University","Sapporo, Hokkaido",null,"https:\u002F\u002Fwww-lmd.ist.hokudai.ac.jp\u002Fmember\u002Fguang-li\u002F","https:\u002F\u002Fgithub.com\u002FGuang000",[85,89,93,97],{"name":86,"color":87,"percentage":88},"HTML","#e34c26",88.4,{"name":90,"color":91,"percentage":92},"CSS","#663399",8.3,{"name":94,"color":95,"percentage":96},"JavaScript","#f1e05a",2.6,{"name":98,"color":99,"percentage":100},"Python","#3572A5",0.6,1919,172,"2026-04-04T05:09:29","MIT",5,"","未说明",{"notes":109,"python":107,"dependencies":110},"该项目是一个数据集蒸馏（Dataset Distillation）领域的论文和资源列表（Awesome List），而非一个可直接运行的单一软件工具。README 中列出了该领域内的数百篇论文及其对应的独立代码仓库链接。具体的运行环境需求（如操作系统、GPU、Python 版本等）取决于用户选择复现的特定论文代码，需参考各子项目的独立文档。",[],[13],[113,114],"awesome-list","deep-learning","2026-03-27T02:49:30.150509","2026-04-06T05:37:35.991140",[118,123,128,133,138,143],{"id":119,"question_zh":120,"answer_zh":121,"source_url":122},9571,"有没有易于使用的数据集蒸馏工具包？","本仓库中提供了许多可用的代码实现。最常用的方法包括 DC (Dataset Condensation)，代码地址：https:\u002F\u002Fgithub.com\u002FVICO-UoE\u002FDatasetCondensation；以及 MTT (Matching Training Trajectories)，代码地址：https:\u002F\u002Fgithub.com\u002Fgeorgecazenavette\u002Fmtt-distillation。您可以点击仓库中的链接查看详细信息。","https:\u002F\u002Fgithub.com\u002FGuang000\u002FAwesome-Dataset-Distillation\u002Fissues\u002F25",{"id":124,"question_zh":125,"answer_zh":126,"source_url":127},9572,"数据集蒸馏技术可以应用于目标检测任务吗？","目前还没有专门针对目标检测任务的数据集蒸馏算法。这是一个具有挑战性但值得尝试的研究方向，如果您打算探索此领域，祝您好运。","https:\u002F\u002Fgithub.com\u002FGuang000\u002FAwesome-Dataset-Distillation\u002Fissues\u002F22",{"id":129,"question_zh":130,"answer_zh":131,"source_url":132},9573,"为什么某些论文（如匿名投稿或特定领域的论文）没有被收录或归类到主要算法部分？","关于收录标准：1. 原则上不收录匿名投稿（如会议评审期间的论文），通常需等待其在 arXiv 公开或被正式录用后，经过社区评估再行添加，以确保列表质量。2. 关于分类，每篇论文通常只归入一个最合适的类别以避免冗余。例如，虽然某篇论文使用了梯度匹配，但如果其核心贡献是针对图数据（GNN）且未在通用视觉基准上证明显著改进，则归类为图数据集蒸馏更为合适。3. 对于使用 GAN 进行图像域蒸馏但思路接近原始论文的作品，可能不被视为主要的新型 DD 算法，但可根据具体内容归入“表格数据”或其他特定子章节。","https:\u002F\u002Fgithub.com\u002FGuang000\u002FAwesome-Dataset-Distillation\u002Fissues\u002F9",{"id":134,"question_zh":135,"answer_zh":136,"source_url":137},9574,"如何申请将自己的新论文添加到该列表中？","您可以在 GitHub 上提交 Issue 提出请求。维护者通常会欢迎高质量的论文加入，并可能将其放入新的相关子章节（如推荐系统、表格数据等）。如果论文涉及特定应用（如医疗、鲁棒性），可能会被归入相应类别。此外，维护者可能会建议或要求作者提供对应的开源代码链接，以便其他研究者复现和使用。","https:\u002F\u002Fgithub.com\u002FGuang000\u002FAwesome-Dataset-Distillation\u002Fissues\u002F41",{"id":139,"question_zh":140,"answer_zh":141,"source_url":142},9575,"列表中标记有代码图标但链接跳转错误的论文，是否有开源代码？","如果遇到标记了代码图标但链接失效或跳转错误（如跳回当前仓库）的情况，这通常是链接配置错误。对于特定的医疗应用论文（如 Soft-Label Anonymous Gastric X-ray Image Distillation），如果作者在 GitHub 上未公开相关代码，则可能暂时无法提供。建议直接联系论文作者索取代码或实现思路，同时也可向仓库维护者反馈以修复错误的链接。","https:\u002F\u002Fgithub.com\u002FGuang000\u002FAwesome-Dataset-Distillation\u002Fissues\u002F8",{"id":144,"question_zh":145,"answer_zh":146,"source_url":147},9576,"如果我提交的论文被拒绝收录或分类不符合预期，原因可能是什么？","主要原因通常包括：1. 论文属于匿名投稿阶段，需等待公开或录用。2. 论文的核心贡献与现有主要算法类别（如通用视觉数据集蒸馏）不完全契合，更适合归入特定领域（如图神经网络、推荐系统、表格数据）。3. 方法论上与已有工作（如原始 GAN 蒸馏论文）差异较小，缺乏在通用基准上的显著改进证据。维护者会根据论文的具体侧重点（如是否针对特定数据结构、是否引入新的正则化项等）来决定最佳分类位置。","https:\u002F\u002Fgithub.com\u002FGuang000\u002FAwesome-Dataset-Distillation\u002Fissues\u002F6",[]]