[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-dennybritz--deeplearning-papernotes":3,"tool-dennybritz--deeplearning-papernotes":65},[4,23,32,40,49,57],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":22},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。",85092,2,"2026-04-10T11:13:16",[13,14,15,16,17,18,19,20,21],"图像","数据工具","视频","插件","Agent","其他","语言模型","开发框架","音频","ready",{"id":24,"name":25,"github_repo":26,"description_zh":27,"stars":28,"difficulty_score":29,"last_commit_at":30,"category_tags":31,"status":22},5784,"funNLP","fighting41love\u002FfunNLP","funNLP 是一个专为中文自然语言处理（NLP）打造的超级资源库，被誉为\"NLP 民工的乐园”。它并非单一的软件工具，而是一个汇集了海量开源项目、数据集、预训练模型和实用代码的综合性平台。\n\n面对中文 NLP 领域资源分散、入门门槛高以及特定场景数据匮乏的痛点，funNLP 提供了“一站式”解决方案。这里不仅涵盖了分词、命名实体识别、情感分析、文本摘要等基础任务的标准工具，还独特地收录了丰富的垂直领域资源，如法律、医疗、金融行业的专用词库与数据集，甚至包含古诗词生成、歌词创作等趣味应用。其核心亮点在于极高的全面性与实用性，从基础的字典词典到前沿的 BERT、GPT-2 模型代码，再到高质量的标注数据和竞赛方案，应有尽有。\n\n无论是刚刚踏入 NLP 领域的学生、需要快速验证想法的算法工程师，还是从事人工智能研究的学者，都能在这里找到急需的“武器弹药”。对于开发者而言，它能大幅减少寻找数据和复现模型的时间；对于研究者，它提供了丰富的基准测试资源和前沿技术参考。funNLP 以开放共享的精神，极大地降低了中文自然语言处理的开发与研究成本，是中文 AI 社区不可或缺的宝藏仓库。",79857,1,"2026-04-08T20:11:31",[19,14,18],{"id":33,"name":34,"github_repo":35,"description_zh":36,"stars":37,"difficulty_score":29,"last_commit_at":38,"category_tags":39,"status":22},5773,"cs-video-courses","Developer-Y\u002Fcs-video-courses","cs-video-courses 是一个精心整理的计算机科学视频课程清单，旨在为自学者提供系统化的学习路径。它汇集了全球知名高校（如加州大学伯克利分校、新南威尔士大学等）的完整课程录像，涵盖从编程基础、数据结构与算法，到操作系统、分布式系统、数据库等核心领域，并深入延伸至人工智能、机器学习、量子计算及区块链等前沿方向。\n\n面对网络上零散且质量参差不齐的教学资源，cs-video-courses 解决了学习者难以找到成体系、高难度大学级别课程的痛点。该项目严格筛选内容，仅收录真正的大学层级课程，排除了碎片化的简短教程或商业广告，确保用户能接触到严谨的学术内容。\n\n这份清单特别适合希望夯实计算机基础的开发者、需要补充特定领域知识的研究人员，以及渴望像在校生一样系统学习计算机科学的自学者。其独特的技术亮点在于分类极其详尽，不仅包含传统的软件工程与网络安全，还细分了生成式 AI、大语言模型、计算生物学等新兴学科，并直接链接至官方视频播放列表，让用户能一站式获取高质量的教育资源，免费享受世界顶尖大学的课堂体验。",79792,"2026-04-08T22:03:59",[18,13,14,20],{"id":41,"name":42,"github_repo":43,"description_zh":44,"stars":45,"difficulty_score":46,"last_commit_at":47,"category_tags":48,"status":22},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。",77062,3,"2026-04-04T04:44:48",[17,13,20,19,18],{"id":50,"name":51,"github_repo":52,"description_zh":53,"stars":54,"difficulty_score":46,"last_commit_at":55,"category_tags":56,"status":22},519,"PaddleOCR","PaddlePaddle\u002FPaddleOCR","PaddleOCR 是一款基于百度飞桨框架开发的高性能开源光学字符识别工具包。它的核心能力是将图片、PDF 等文档中的文字提取出来，转换成计算机可读取的结构化数据，让机器真正“看懂”图文内容。\n\n面对海量纸质或电子文档，PaddleOCR 解决了人工录入效率低、数字化成本高的问题。尤其在人工智能领域，它扮演着连接图像与大型语言模型（LLM）的桥梁角色，能将视觉信息直接转化为文本输入，助力智能问答、文档分析等应用场景落地。\n\nPaddleOCR 适合开发者、算法研究人员以及有文档自动化需求的普通用户。其技术优势十分明显：不仅支持全球 100 多种语言的识别，还能在 Windows、Linux、macOS 等多个系统上运行，并灵活适配 CPU、GPU、NPU 等各类硬件。作为一个轻量级且社区活跃的开源项目，PaddleOCR 既能满足快速集成的需求，也能支撑前沿的视觉语言研究，是处理文字识别任务的理想选择。",75489,"2026-04-13T11:13:28",[19,13,20,18],{"id":58,"name":59,"github_repo":60,"description_zh":61,"stars":62,"difficulty_score":29,"last_commit_at":63,"category_tags":64,"status":22},3215,"awesome-machine-learning","josephmisiti\u002Fawesome-machine-learning","awesome-machine-learning 是一份精心整理的机器学习资源清单，汇集了全球优秀的机器学习框架、库和软件工具。面对机器学习领域技术迭代快、资源分散且难以甄选的痛点，这份清单按编程语言（如 Python、C++、Go 等）和应用场景（如计算机视觉、自然语言处理、深度学习等）进行了系统化分类，帮助使用者快速定位高质量项目。\n\n它特别适合开发者、数据科学家及研究人员使用。无论是初学者寻找入门库，还是资深工程师对比不同语言的技术选型，都能从中获得极具价值的参考。此外，清单还延伸提供了免费书籍、在线课程、行业会议、技术博客及线下聚会等丰富资源，构建了从学习到实践的全链路支持体系。\n\n其独特亮点在于严格的维护标准：明确标记已停止维护或长期未更新的项目，确保推荐内容的时效性与可靠性。作为机器学习领域的“导航图”，awesome-machine-learning 以开源协作的方式持续更新，旨在降低技术探索门槛，让每一位从业者都能高效地站在巨人的肩膀上创新。",72149,"2026-04-03T21:50:24",[20,18],{"id":66,"github_repo":67,"name":68,"description_en":69,"description_zh":70,"ai_summary_zh":70,"readme_en":71,"readme_zh":72,"quickstart_zh":73,"use_case_zh":74,"hero_image_url":75,"owner_login":76,"owner_name":77,"owner_avatar_url":78,"owner_bio":79,"owner_company":79,"owner_location":80,"owner_email":79,"owner_twitter":79,"owner_website":81,"owner_url":82,"languages":79,"stars":83,"forks":84,"last_commit_at":85,"license":79,"difficulty_score":29,"env_os":86,"env_gpu":87,"env_ram":87,"env_deps":88,"category_tags":91,"github_topics":79,"view_count":10,"oss_zip_url":79,"oss_zip_packed_at":79,"status":22,"created_at":92,"updated_at":93,"faqs":94,"releases":95},7147,"dennybritz\u002Fdeeplearning-papernotes","deeplearning-papernotes","Summaries and notes on Deep Learning research papers","deeplearning-papernotes 是一个专注于深度学习领域的开源知识库，旨在为研究者提供高质量的研究论文摘要与笔记。面对深度学习领域论文爆发式增长、阅读门槛高且难以快速把握核心思想的痛点，该项目通过人工梳理，将复杂的学术成果转化为结构清晰、易于理解的精华内容。\n\n它特别适合人工智能研究人员、算法工程师以及希望紧跟前沿技术的学生使用。无论是需要快速调研某个细分方向（如强化学习、计算机视觉或自然语言处理），还是希望深入理解矩阵微积分等基础理论，用户都能在此找到按时间线整理的精选论文解读。从 IMPALA 分布式架构到 DensePose 人体姿态估计，再到对抗样本与可解释性分析，内容覆盖广泛且更新及时。\n\n其独特亮点在于不仅提供论文链接，还往往附带官方代码库、技术博客文章及相关资源索引，帮助用户实现从理论理解到代码复现的无缝衔接。通过社区协作的方式，deeplearning-papernotes 持续沉淀有价值的学术洞察，是深度学习从业者高效获取知识、节省文献筛选时间的得力助手。","#### 2018-02\n\n- The Matrix Calculus You Need For Deep Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1802.01528v2)]\n- Regularized Evolution for Image Classifier Architecture Search [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1802.01548)]\n- Online Learning: A Comprehensive Survey [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1802.02871)]\n- Visual Interpretability for Deep Learning: a Survey [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1802.00614)]\n- Behavior is Everything – Towards Representing Concepts with Sensorimotor Contingencies [[paper](https:\u002F\u002Fwww.vicarious.com\u002Fwp-content\u002Fuploads\u002F2018\u002F01\u002FAAAI18-pixelworld.pdf)] [[article](https:\u002F\u002Fwww.vicarious.com\u002F2018\u002F02\u002F07\u002Flearning-concepts-through-sensorimotor-interactions\u002F)] [[code](https:\u002F\u002Fgithub.com\u002Fvicariousinc\u002Fpixelworld)]\n- IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1802.01561)] [[article](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fimpala-scalable-distributed-deeprl-dmlab-30\u002F)] [[code](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Flab\u002Ftree\u002Fmaster\u002Fgame_scripts\u002Flevels\u002Fcontributed\u002Fdmlab30)]\n- DeepType: Multilingual Entity Linking by Neural Type System Evolution [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1802.01021)] [[article](https:\u002F\u002Fblog.openai.com\u002Fdiscovering-types-for-entity-disambiguation\u002F)] [[code](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fdeeptype)]\n- DensePose: Dense Human Pose Estimation In The Wild [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1802.00434)] [[article](http:\u002F\u002Fdensepose.org\u002F)]\n\n#### 2018-01\n\n- Nested LSTMs [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.10308)]\n- Generating Wikipedia by Summarizing Long Sequences [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.10198)]\n- Scalable and accurate deep learning for electronic health records [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.07860)]\n- Kernel Feature Selection via Conditional Covariance Minimization [[NIPS paper](https:\u002F\u002Fpapers.nips.cc\u002Fpaper\u002F7270-kernel-feature-selection-via-conditional-covariance-minimization.pdf)] [[article](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2018\u002F01\u002F23\u002Fkernels\u002F)] [[code](https:\u002F\u002Fgithub.com\u002FJianbo-Lab\u002FCCM)]\n- Psychlab: A Psychology Laboratory for Deep Reinforcement Learning Agents [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.08116)] [[article](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fopen-sourcing-psychlab\u002F)] [[code](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Flab\u002Ftree\u002Fmaster\u002Fgame_scripts\u002Flevels\u002Fcontributed\u002Fpsychlab)]\n- Fine-tuned Language Models for Text Classification [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.06146)] [[code]()] (soon)\n- Deep Learning: An Introduction for Applied Mathematicians [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.05894v1)]\n- Innateness, AlphaZero, and Artificial Intelligence [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.05667)]\n- Can Computers Create Art? [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.04486)]\n- eCommerceGAN : A Generative Adversarial Network for E-commerce [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.03244)]\n- Expected Policy Gradients for Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.03326)]\n- DroNet: Learning to Fly by Driving [[UZH docs](http:\u002F\u002Frpg.ifi.uzh.ch\u002Fdocs\u002FRAL18_Loquercio.pdf)] [[article](http:\u002F\u002Frpg.ifi.uzh.ch\u002Fdronet.html)] [[code](https:\u002F\u002Fgithub.com\u002Fuzh-rpg\u002Frpg_public_dronet)]\n- Symmetric Decomposition of Asymmetric Games [[Scientific Reports](https:\u002F\u002Fwww.nature.com\u002Farticles\u002Fs41598-018-19194-4)] [[article](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fgame-theory-insights-asymmetric-multi-agent-games\u002F)]\n- Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.01290)] [[code](https:\u002F\u002Fgithub.com\u002Fhaarnoja\u002Fsac)]\n- SBNet: Sparse Blocks Network for Fast Inference [[arXiv](https:\u002F\u002Farxiv.org\u002Fpdf\u002F1801.02108.pdf)] [[article](https:\u002F\u002Feng.uber.com\u002Fsbnet\u002F)] [[code](https:\u002F\u002Fgithub.com\u002Fuber\u002Fsbnet)]\n- DeepMind Control Suite [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.00690)] [[code](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Fdm_control)]\n- Deep Learning: A Critical Appraisal [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.00631)]\n\n\n#### 2017-12\n\n- Adversarial Patch [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.09665)]\n- CNN Is All You Need [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.09662)]\n- Learning Robot Objectives from Physical Human Interaction [[paper](http:\u002F\u002Fproceedings.mlr.press\u002Fv78\u002Fbajcsy17a\u002Fbajcsy17a.pdf)] [[article](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2018\u002F02\u002F06\u002Fphri\u002F)]\n- The NarrativeQA Reading Comprehension Challenge [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.07040v1)] [[dataset](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Fnarrativeqa)]\n- Objects that Sound [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.06651)]\n- Natural TTS Synthesis by Conditioning WaveNet on Mel Spectrogram Predictions [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.05884)] [[article](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F12\u002Ftacotron-2-generating-human-like-speech.html)] [[article2](https:\u002F\u002Fgoogle.github.io\u002Ftacotron\u002Fpublications\u002Ftacotron2\u002Findex.html)]\n- Deep Neuroevolution: Genetic Algorithms Are a Competitive Alternative for Training Deep Neural Networks for Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.06567)] [[article](https:\u002F\u002Feng.uber.com\u002Fdeep-neuroevolution\u002F)] [[code](https:\u002F\u002Fgithub.com\u002Fuber-common\u002Fdeep-neuroevolution)]\n- Improving Exploration in Evolution Strategies for Deep Reinforcement Learning via a Population of Novelty-Seeking Agents [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.06560)] [[article](https:\u002F\u002Feng.uber.com\u002Fdeep-neuroevolution\u002F)] [[code](https:\u002F\u002Fgithub.com\u002Fuber-common\u002Fdeep-neuroevolution)]\n- Superhuman AI for heads-up no-limit poker: Libratus beats top professionals [[Science](http:\u002F\u002Fscience.sciencemag.org\u002Fcontent\u002Fearly\u002F2017\u002F12\u002F15\u002Fscience.aao1733)]\n- Mathematics of Deep Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.04741)]\n- State-of-the-art Speech Recognition With Sequence-to-Sequence Models [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.01769)] [[article](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F12\u002Fimproving-end-to-end-models-for-speech.html)]\n- Peephole: Predicting Network Performance Before Training [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.03351)]\n- Deliberation Network: Pushing the frontiers of neural machine translation [[Research at Microsoft](https:\u002F\u002Fwww.microsoft.com\u002Fen-us\u002Fresearch\u002Fpublication\u002Fdeliberation-networks-sequence-generation-beyond-one-pass-decoding\u002F)] [[article](https:\u002F\u002Fwww.microsoft.com\u002Fen-us\u002Fresearch\u002Fblog\u002Fdeliberation-networks\u002F)]\n- GPU Kernels for Block-Sparse Weights [[Research at OpenAI](https:\u002F\u002Fs3-us-west-2.amazonaws.com\u002Fopenai-assets\u002Fblocksparse\u002Fblocksparsepaper.pdf)] [[article](https:\u002F\u002Fblog.openai.com\u002Fblock-sparse-gpu-kernels\u002F)] [[code](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fblocksparse)]\n- Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithm [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.01815)]\n- Deep Learning Scaling is Predictable, Empirically [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.00409)] [[article](http:\u002F\u002Fresearch.baidu.com\u002Fdeep-learning-scaling-predictable-empirically\u002F)]\n\n#### 2017-11\n\n- High-Resolution Image Synthesis and Semantic Manipulation with Conditional GANs [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.11585)] [[article](https:\u002F\u002Ftcwang0509.github.io\u002Fpix2pixHD\u002F)] [[code](https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fpix2pixHD)]\n- StarGAN: Unified Generative Adversarial Networks for Multi-Domain Image-to-Image Translation [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.09020)] [[code](https:\u002F\u002Fgithub.com\u002Fyunjey\u002FStarGAN\u002F)]\n- Population Based Training of Neural Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.09846)] [[article](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fpopulation-based-training-neural-networks\u002F)]\n- Distilling a Neural Network Into a Soft Decision Tree [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.09784)]\n- Neural Text Generation: A Practical Guide [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.09534)]\n- Parallel WaveNet: Fast High-Fidelity Speech Synthesis [[DeepMind documents](https:\u002F\u002Fdeepmind.com\u002Fdocuments\u002F131\u002FDistilling_WaveNet.pdf)] [[article](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fhigh-fidelity-speech-synthesis-wavenet\u002F)]\n- CheXNet: Radiologist-Level Pneumonia Detection on Chest X-Rays with Deep Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.05225)] [[article](https:\u002F\u002Fstanfordmlgroup.github.io\u002Fprojects\u002Fchexnet\u002F)]\n- Non-local Neural Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.07971)]\n- Deep Image Prior [[paper](https:\u002F\u002Fsites.skoltech.ru\u002Fapp\u002Fdata\u002Fuploads\u002Fsites\u002F25\u002F2017\u002F11\u002Fdeep_image_prior.pdf)] [[article](https:\u002F\u002Fdmitryulyanov.github.io\u002Fdeep_image_prior)] [[code](https:\u002F\u002Fgithub.com\u002FDmitryUlyanov\u002Fdeep-image-prior)]\n- Online Deep Learning: Learning Deep Neural Networks on the Fly [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.03705)]\n- Learning Explanatory Rules from Noisy Data [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.04574)]\n- Improving Palliative Care with Deep Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.06402)] [[article](https:\u002F\u002Fstanfordmlgroup.github.io\u002Fprojects\u002Fimproving-palliative-care\u002F)]\n- VoxelNet: End-to-End Learning for Point Cloud Based 3D Object Detection [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.06396)]\n- Weighted Transformer Network for Machine Translation [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.02132)] [[article](https:\u002F\u002Feinstein.ai\u002Fresearch\u002Fweighted-transformer)]\n- Non-Autoregressive Neural Machine Translation [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.02281)] [[article](https:\u002F\u002Feinstein.ai\u002Fresearch\u002Fnon-autoregressive-neural-machine-translation)]\n- Block-Sparse Recurrent Neural Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.02782)]\n- A Unified Game-Theoretic Approach to Multiagent Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.00832)]\n- Neural Discrete Representation Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.00937)] [[article](https:\u002F\u002Favdnoord.github.io\u002Fhomepage\u002Fvqvae\u002F)]\n- Don't Decay the Learning Rate, Increase the Batch Size [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.00489)]\n- Hierarchical Representations for Efficient Architecture Search [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.00436)]\n\n#### 2017-10\n\n- Unsupervised Machine Translation Using Monolingual Corpora Only [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.00043)]\n- Dynamic Routing Between Capsules [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.09829)]\n- A generative vision model that trains with high data efficiency and breaks text-based CAPTCHAs [[Science](http:\u002F\u002Fscience.sciencemag.org\u002Fcontent\u002Fearly\u002F2017\u002F10\u002F26\u002Fscience.aag2612.full)] [[article](https:\u002F\u002Fwww.vicarious.com\u002F2017\u002F10\u002F26\u002Fcommon-sense-cortex-and-captcha\u002F)] [[code](https:\u002F\u002Fgithub.com\u002Fvicariousinc\u002Fscience_rcn)]\n- Understanding Grounded Language Learning Agents [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.09867)]\n- Planning, Fast and Slow: A Framework for Adaptive Real-Time Safe Trajectory Planning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.04731)] [[article](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2017\u002F12\u002F05\u002Ffastrack\u002F)] [[code](https:\u002F\u002Fgithub.com\u002FHJReachability)] (soon)\n- Malware Detection by Eating a Whole EXE [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.09435)] [[article](https:\u002F\u002Fdevblogs.nvidia.com\u002Fmalware-detection-neural-networks\u002F)]\n- Progressive Growing of GANs for Improved Quality, Stability, and Variation [[Research at Nvidia](http:\u002F\u002Fresearch.nvidia.com\u002Fsites\u002Fdefault\u002Ffiles\u002Fpubs\u002F2017-10_Progressive-Growing-of\u002F\u002Fkarras2017gan-paper.pdf)] [[article](http:\u002F\u002Fresearch.nvidia.com\u002Fpublication\u002F2017-10_Progressive-Growing-of)] [[code](https:\u002F\u002Fgithub.com\u002Ftkarras\u002Fprogressive_growing_of_gans)]\n- Meta Learning Shared Hierarchies [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.09767)] [[article](https:\u002F\u002Fblog.openai.com\u002Flearning-a-hierarchy\u002F)] [[code](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fmlsh)]\n- Deep Voice 3: 2000-Speaker Neural Text-to-Speech [[arXiv](http:\u002F\u002Fresearch.baidu.com\u002Fdeep-voice-3-2000-speaker-neural-text-speech\u002F)] [[article](http:\u002F\u002Fresearch.baidu.com\u002Fdeep-voice-3-2000-speaker-neural-text-speech\u002F)]\n- AVA: A Video Dataset of Spatio-temporally Localized Atomic Visual Actions [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.08421)] [[article](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F10\u002Fannouncing-ava-finely-labeled-video.html)] [[dataset](https:\u002F\u002Fresearch.google.com\u002Fava\u002F)]\n-  Mastering the game of Go without Human Knowledge [[Nature](https:\u002F\u002Fwww.nature.com\u002Farticles\u002Fnature24270.epdf?author_access_token=VJXbVjaSHxFoctQQ4p2k4tRgN0jAjWel9jnR3ZoTv0PVW4gB86EEpGqTRDtpIz-2rmo8-KG06gqVobU5NSCFeHILHcVFUeMsbvwS-lxjqQGg98faovwjxeTUgZAUMnRQ)] [[article](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Falphago-zero-learning-scratch\u002F)]\n-  Sim-to-Real Transfer of Robotic Control with Dynamics Randomization [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.06537)] [[article](https:\u002F\u002Fblog.openai.com\u002Fgeneralizing-from-simulation\u002F)]\n-  Asymmetric Actor Critic for Image-Based Robot Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.06542)] [[article](https:\u002F\u002Fblog.openai.com\u002Fgeneralizing-from-simulation\u002F)]\n-  A systematic study of the class imbalance problem in convolutional neural networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.05381)]\n-  Generalization in Deep Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.05468)]\n- Swish: a Self-Gated Activation Function [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.05941)]\n- Emergent Translation in Multi-Agent Communication [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.06922)]\n- SLING: A framework for frame semantic parsing [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.07032)] [[article](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F11\u002Fsling-natural-language-frame-semantic.html)] [[code](https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fsling)]\n- Meta-Learning for Wrestling [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.03641)] [[article](https:\u002F\u002Fblog.openai.com\u002Fmeta-learning-for-wrestling\u002F)] [[code](https:\u002F\u002Fgithub.com\u002Fopenai\u002Frobosumo)]\n- Mixed Precision Training [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.03740)] [[article](http:\u002F\u002Fresearch.baidu.com\u002Fmixed-precision-training\u002F)] [[article2](https:\u002F\u002Fdevblogs.nvidia.com\u002Fparallelforall\u002Fmixed-precision-training-deep-neural-networks\u002F)] [[code\u002Fdocs](http:\u002F\u002Fdocs.nvidia.com\u002Fdeeplearning\u002Fsdk\u002Fmixed-precision-training\u002Findex.html)]\n- Generative Adversarial Networks: An Overview [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.07035)]\n- Emergent Complexity via Multi-Agent Competition [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.03748)] [[article](https:\u002F\u002Fblog.openai.com\u002Fcompetitive-self-play\u002F)] [[code](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fmultiagent-competition)]\n- Deep Lattice Networks and Partial Monotonic Functions [[Research at Google](https:\u002F\u002Fresearch.google.com\u002Fpubs\u002Fpub46327.html)] [[article](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F10\u002Ftensorflow-lattice-flexibility.html)] [[code](https:\u002F\u002Fgithub.com\u002Ftensorflow\u002Flattice)]\n- The IIT Bombay English-Hindi Parallel Corpus [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.02855)] [[article](http:\u002F\u002Fwww.cfilt.iitb.ac.in\u002Fiitb_parallel\u002F)]\n- Rainbow: Combining Improvements in Deep Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.02298)]\n- Lifelong Learning With Dynamically Expandable Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.01547)]\n- Variational Inference & Deep Learning: A New Synthesis (Thesis) [[dropbox](https:\u002F\u002Fwww.dropbox.com\u002Fs\u002Fv6ua3d9yt44vgb3\u002Fcover_and_thesis.pdf)]\n- Neural Task Programming: Learning to Generalize Across Hierarchical Tasks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.01813)]\n- Neural Color Transfer between Images [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.00756)]\n- The hippocampus as a predictive map [[biorXiv](https:\u002F\u002Fwww.biorxiv.org\u002Fcontent\u002Fbiorxiv\u002Fearly\u002F2017\u002F07\u002F25\u002F097170.full.pdf)] [[article](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fhippocampus-predictive-map\u002F)]\n- Scalable and accurate deep learning for electronic health\nrecords [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.07860)]\n\n#### 2017-09\n\n- Variational Memory Addressing in Generative Models [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.07116)]\n- Overcoming Exploration in Reinforcement Learning with Demonstrations [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.10089)]\n- A Hybrid DSP\u002FDeep Learning Approach to Real-Time Full-Band Speech Enhancement [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.08243)] [[article](https:\u002F\u002Fpeople.xiph.org\u002F~jm\u002Fdemo\u002Frnnoise\u002F)] [[code](https:\u002F\u002Fgithub.com\u002Fxiph\u002Frnnoise\u002F)]\n- ChestX-ray8: Hospital-scale Chest X-ray Database and Benchmarks on\nWeakly-Supervised Classification and Localization of Common Thorax Diseases [[CVF](http:\u002F\u002Fopenaccess.thecvf.com\u002Fcontent_cvpr_2017\u002Fpapers\u002FWang_ChestX-ray8_Hospital-Scale_Chest_CVPR_2017_paper.pdf)] [[article](https:\u002F\u002Fwww.nih.gov\u002Fnews-events\u002Fnews-releases\u002Fnih-clinical-center-provides-one-largest-publicly-available-chest-x-ray-datasets-scientific-community)] [[dataset](https:\u002F\u002Fnihcc.app.box.com\u002Fv\u002FChestXray-NIHCC)]\n- NIMA: Neural Image Assessment [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.05424)] [[article](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F12\u002Fintroducing-nima-neural-image-assessment.html)]\n- Generating Sentences by Editing Prototypes [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.08878)] [[code](https:\u002F\u002Fgithub.com\u002Fkelvinguu\u002Fneural-editor)]\n- The Consciousness Prior [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.08568)]\n- StarSpace: Embed All The Things! [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.03856)] [[code](https:\u002F\u002Fgithub.com\u002Ffacebookresearch\u002FStarspace)]\n- Neural Optimizer Search with Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.07417)]\n- Dynamic Evaluation of Neural Sequence Models [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.07432)]\n- Neural Machine Translation [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.07809)]\n- Matterport3D: Learning from RGB-D Data in Indoor Environments [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.06158)] [[article](https:\u002F\u002Fniessner.github.io\u002FMatterport\u002F)] [[article2](https:\u002F\u002Fhackernoon.com\u002Fannouncing-the-matterport3d-research-dataset-815cae932939)] [[code](https:\u002F\u002Fgithub.com\u002Fniessner\u002FMatterport)]\n- Deep Reinforcement Learning that Matters [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.06560)] [[code](https:\u002F\u002Fgithub.com\u002FBreakend\u002FDeepReinforcementLearningThatMatters)]\n- The Uncertainty Bellman Equation and Exploration [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.05380)]\n- WESPE: Weakly Supervised Photo Enhancer for Digital Cameras [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.01118)] [[article](http:\u002F\u002Fpeople.ee.ethz.ch\u002F~ihnatova\u002Fwespe.html)]\n- Globally Normalized Reader [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.02828)] [[article](http:\u002F\u002Fresearch.baidu.com\u002Fgnr\u002F)] [[code](https:\u002F\u002Fgithub.com\u002Fbaidu-research\u002FGloballyNormalizedReader)]\n- A Brief Introduction to Machine Learning for Engineers [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.02840)]\n- Learning with Opponent-Learning Awareness [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.04326)] [[article](https:\u002F\u002Fblog.openai.com\u002Flearning-to-model-other-minds\u002F)]\n- A Deep Reinforcement Learning Chatbot [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.02349)]\n- Squeeze-and-Excitation Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.01507)]\n- Efficient Methods and Hardware for Deep Learning (Thesis) [[Stanford Digital Repository](https:\u002F\u002Fpurl.stanford.edu\u002Fqf934gh3708)]\n\n#### 2017-08\n\n- Design and Analysis of the NIPS 2016 Review Process [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.09794)]\n- Fast Automated Analysis of Strong Gravitational Lenses with Convolutional Neural Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.08842)] [[article](http:\u002F\u002Fwww.symmetrymagazine.org\u002Farticle\u002Fneural-networks-meet-space)]\n- TensorFlow Agents: Efficient Batched Reinforcement Learning in TensorFlow [[white paper](https:\u002F\u002Fdrive.google.com\u002Ffile\u002Fd\u002F0B20Yn-GSaVHGMVlPanRTRlNIRlk\u002Fview)] [[code](https:\u002F\u002Fgithub.com\u002Ftensorflow\u002Fagents)]\n- Automated Crowdturfing Attacks and Defenses in Online Review Systems [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.08151)]\n- Neural Network Dynamics for Model-Based Deep Reinforcement Learning with Model-Free Fine-Tuning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.02596)] [[article](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2017\u002F11\u002F30\u002Fmodel-based-rl\u002F)] [[code](https:\u002F\u002Fgithub.com\u002Fnagaban2\u002Fnn_dynamics)]\n- Deep Learning for Video Game Playing [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.07902)]\n- Deep & Cross Network for Ad Click Predictions [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.05123)]\n- Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithms [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.07747)] [[code](https:\u002F\u002Fgithub.com\u002Fzalandoresearch\u002Ffashion-mnist)]\n- Multi-task Self-Supervised Visual Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.07860)]\n- Learning a Multi-View Stereo Machine [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.05375)] [[article](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2017\u002F09\u002F05\u002Funified-3d\u002F)] [[code]()] (soon)\n- Twin Networks: Using the Future as a Regularizer [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.06742)]\n- A Brief Survey of Deep Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.05866)]\n- Scalable trust-region method for deep reinforcement learning using Kronecker-factored approximation [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.05144)] [[code](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fbaselines)]\n- On the Effectiveness of Visible Watermarks [[CVPR](http:\u002F\u002Fopenaccess.thecvf.com\u002Fcontent_cvpr_2017\u002Fpapers\u002FDekel_On_the_Effectiveness_CVPR_2017_paper.pdf)] [[article](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F08\u002Fmaking-visible-watermarks-more-effective.html)]\n- Practical Network Blocks Design with Q-Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.05552)]\n- On Ensuring that Intelligent Machines Are Well-Behaved [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.05448)]\n- Reproducibility of Benchmarked Deep Reinforcement Learning Tasks for Continuous Control [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.04133)] [[code](https:\u002F\u002Fgithub.com\u002FBreakend\u002FReproducibilityInContinuousPolicyGradientMethods)]\n- Training Deep AutoEncoders for Collaborative Filtering [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.01715)] [[code](https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FDeepRecommender)]\n- Learning to Perform a Perched Landing on the GroundUsing Deep Reinforcement Learning [[nature](https:\u002F\u002Flink.springer.com\u002Fepdf\u002F10.1007\u002Fs10846-017-0696-1?author_access_token=BEvJgzY3QauUddBuQAus2ve4RwlQNchNByi7wbcMAY5xhRRqI6HVNnXt8Pgp850SnuV5ue6mUo3Jc7FIP5FgLmqk34Wob3oqyuGtkg7E_1T0dg02IYhfY-3dvb8R9zEmaGzTogYCIXm4O4vZ_tSGnA%3D%3D)]\n- Revisiting the Effectiveness of Off-the-shelf Temporal Modeling Approaches for Large-scale Video Classification [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.03805)] [[article](http:\u002F\u002Fresearch.baidu.com\u002Fspatial-temporal-modeling-framework-large-scale-video-understanding\u002F)]\n- Intrinsically Motivated Goal Exploration Processes with Automatic Curriculum Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.02190)]\n- Neural Expectation Maximization [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.03498)] [[code](https:\u002F\u002Fgithub.com\u002Fsjoerdvansteenkiste\u002F)]\n- Google Vizier: A Service for Black-Box Optimization [[Research at Google](https:\u002F\u002Fresearch.google.com\u002Fpubs\u002Fpub46180.html)]\n- STARDATA: A StarCraft AI Research Dataset [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.02139)] [[code](https:\u002F\u002Fgithub.com\u002FTorchCraft\u002FStarData)]\n- Using millions of emoji occurrences to learn any-domain representations for detecting sentiment, emotion and sarcasm [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.00524)] [[code](https:\u002F\u002Fgithub.com\u002Fbfelbo\u002Fdeepmoji)] [[article](https:\u002F\u002Fwww.media.mit.edu\u002Fposts\u002Fwhat-can-we-learn-from-emojis\u002F)]\n- Natural Language Processing with Small Feed-Forward Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.00214)]\n\n#### 2017-07\n\n- Photographic Image Synthesis with Cascaded Refinement Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.09405)] [[code](https:\u002F\u002Fgithub.com\u002FCQFIO\u002FPhotographicImageSynthesis)]\n- StarCraft II: A New Challenge for Reinforcement Learning [[DeepMind Documents](https:\u002F\u002Fdeepmind.com\u002Fdocuments\u002F110\u002Fsc2le.pdf)] [[code](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Fpysc2)] [[article](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fdeepmind-and-blizzard-open-starcraft-ii-ai-research-environment\u002F)]\n- Leveraging Demonstrations for Deep Reinforcement Learning on Robotics Problems with Sparse Rewards [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.08817)]\n- Reinforcement Learning with Deep Energy-Based Policies [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1702.08165)] [[article](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2017\u002F10\u002F06\u002Fsoft-q-learning\u002F)] [[code](https:\u002F\u002Fgithub.com\u002Fhaarnoja\u002Fsoftqlearning)]\n- DARLA: Improving Zero-Shot Transfer in Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.08475)]\n- Synthesizing Robust Adversarial Examples [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.07397)] [[article](http:\u002F\u002Fwww.labsix.org\u002Fphysical-objects-that-fool-neural-nets\u002F)] [[code]()] (Soon)\n- Voice Synthesis for in-the-Wild Speakers via a Phonological Loop [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.06588)] [[code](https:\u002F\u002Fgithub.com\u002Ffacebookresearch\u002Floop)] [[article](https:\u002F\u002Fytaigman.github.io\u002Floop\u002F)]\n- Eyemotion: Classifying facial expressions in VR using eye-tracking cameras [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.07204)] [[article](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F07\u002Fexpressions-in-virtual-reality.html)]\n- A Distributional Perspective on Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.06887)] [[article](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fgoing-beyond-average-reinforcement-learning\u002F)] [[video](https:\u002F\u002Fvimeo.com\u002F235922311)]\n- On the State of the Art of Evaluation in Neural Language Models [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.05589)]\n- Optimizing the Latent Space of Generative Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.05776)]\n- Neuroscience-Inspired Artificial Intelligence [[Neuron](http:\u002F\u002Fwww.cell.com\u002Fneuron\u002Ffulltext\u002FS0896-6273(17)30509-3?_returnURL=http%3A%2F%2Flinkinghub.elsevier.com%2Fretrieve%2Fpii%2FS0896627317305093%3Fshowall%3Dtrue)] [[article](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fai-and-neuroscience-virtuous-circle\u002F)]\n- Learning Transferable Architectures for Scalable Image Recognition [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.07012)]\n- Reverse Curriculum Generation for Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.05300)]\n- Imagination-Augmented Agents for Deep Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.06203)] [[article](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fagents-imagine-and-plan\u002F)]\n- Learning model-based planning from scratch [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.06170)] [[article](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fagents-imagine-and-plan\u002F)]\n- Proximal Policy Optimization Algorithms [[AWSS3](https:\u002F\u002Fopenai-public.s3-us-west-2.amazonaws.com\u002Fblog\u002F2017-07\u002Fppo\u002Fppo-arxiv.pdf)] [[code](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fbaselines)]\n- Automatic Recognition of Deceptive Facial Expressions of Emotion [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.04061)]\n- Distral: Robust Multitask Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.04175)]\n- Creatism: A deep-learning photographer capable of creating professional work [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.03491)] [[article](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F07\u002Fusing-deep-learning-to-create.html)]\n- SCAN: Learning Abstract Hierarchical Compositional Visual Concepts [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.03389)] [[article](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fimagine-creating-new-visual-concepts-recombining-familiar-ones\u002F)]\n- Revisiting Unreasonable Effectiveness of Data in Deep Learning Era [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.02968)] [[article](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F07\u002Frevisiting-unreasonable-effectiveness.html)]\n- The Intentional Unintentional Agent: Learning to Solve Many Continuous Control Tasks Simultaneously [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.03300)]\n- Deep Bilateral Learning for Real-Time Image Enhancement [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.02880)] [[code](https:\u002F\u002Fgithub.com\u002Fmgharbi\u002Fhdrnet)] [[article](https:\u002F\u002Fgroups.csail.mit.edu\u002Fgraphics\u002Fhdrnet\u002F)]\n- Emergence of Locomotion Behaviours in Rich Environments [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.02286)] [[article](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fproducing-flexible-behaviours-simulated-environments\u002F)]\n- Learning human behaviors from motion capture by adversarial imitation [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.02201)] [[article](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fproducing-flexible-behaviours-simulated-environments\u002F)]\n- Robust Imitation of Diverse Behaviors [[arXiv](https:\u002F\u002Fdeepmind.com\u002Fdocuments\u002F95\u002Fdiverse_arxiv.pdf)] [[article](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fproducing-flexible-behaviours-simulated-environments\u002F)]\n- [Hindsight Experience Replay](notes\u002Fhindsight-ep.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.01495)]\n- Cardiologist-Level Arrhythmia Detection with Convolutional Neural Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.01836)] [[article](https:\u002F\u002Fstanfordmlgroup.github.io\u002Fprojects\u002Fecg\u002F)]\n- End-to-End Learning of Semantic Grasping [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.01932)]\n- ELF: An Extensive, Lightweight and Flexible Research Platform for Real-time Strategy Games [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.01067)] [[code](https:\u002F\u002Fgithub.com\u002Ffacebookresearch\u002FELF)] [[article](https:\u002F\u002Fcode.facebook.com\u002Fposts\u002F132985767285406\u002Fintroducing-elf-an-extensive-lightweight-and-flexible-platform-for-game-research\u002F)]\n\n#### 2017-06\n\n- [Noisy Networks for Exploration](notes\u002Fnoisy-networks-4-exploration.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.10295)]\n- Do GANs actually learn the distribution? An empirical study [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.08224)]\n- Gradient Episodic Memory for Continuum Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.08840)]\n- Natural Language Does Not Emerge 'Naturally' in Multi-Agent Dialog [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.08502)] [[code](https:\u002F\u002Fgithub.com\u002Fbatra-mlp-lab\u002Flang-emerge)]\n- Deep Interest Network for Click-Through Rate Prediction [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.06978)]\n- Cognitive Psychology for Deep Neural Networks: A Shape Bias Case Study [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.08606)] [[article](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fcognitive-psychology\u002F)]\n- Structure Learning in Motor Control: A Deep Reinforcement Learning Model [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.06827)]\n- Programmable Agents [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.06383)]\n- Grounded Language Learning in a Simulated 3D World [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.06551)]\n- Schema Networks: Zero-shot Transfer with a Generative Causal Model of Intuitive Physics [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.04317)]\n- SVCCA: Singular Vector Canonical Correlation Analysis for Deep Learning Dynamics and Interpretability [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.05806)] [[article](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F11\u002Finterpreting-deep-neural-networks-with.html)] [[code](https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fsvcca)]\n- One Model To Learn Them All [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.05137)] [[code](https:\u002F\u002Fgithub.com\u002Ftensorflow\u002Ftensor2tensor)] [[article](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F06\u002Fmultimodel-multi-task-machine-learning.html)]\n- Hybrid Reward Architecture for Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.04208)]\n- Expected Policy Gradients [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.05374)]\n- Variational Approaches for Auto-Encoding Generative Adversarial Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.04987)]\n- Deal or No Deal? End-to-End Learning for Negotiation Dialogues [[S3AWS](https:\u002F\u002Fs3.amazonaws.com\u002Fend-to-end-negotiator\u002Fend-to-end-negotiator.pdf)] [[code](https:\u002F\u002Fgithub.com\u002Ffacebookresearch\u002Fend-to-end-negotiator)] [[article](https:\u002F\u002Fcode.facebook.com\u002Fposts\u002F1686672014972296\u002Fdeal-or-no-deal-training-ai-bots-to-negotiate\u002F)]\n- Attention Is All You Need [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.03762)] [[code](https:\u002F\u002Fgithub.com\u002Ftensorflow\u002Ftensor2tensor)] [[article](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F08\u002Ftransformer-novel-neural-network.html)]\n- Sobolev Training for Neural Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.04859)]\n- YellowFin and the Art of Momentum Tuning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.03471)] [[code](https:\u002F\u002Fgithub.com\u002FJianGoForIt\u002FYellowFin)] [[article](http:\u002F\u002Fdawn.cs.stanford.edu\u002F2017\u002F07\u002F05\u002Fyellowfin\u002F)]\n- Forward Thinking: Building and Training Neural Networks One Layer at a Time [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.02480)]\n- Depthwise Separable Convolutions for Neural Machine Translation [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.03059)] [[code](https:\u002F\u002Fgithub.com\u002Ftensorflow\u002Ftensor2tensor)]\n- Parameter Space Noise for Exploration [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.01905)] [[code](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fbaselines)] [[article](https:\u002F\u002Fblog.openai.com\u002Fbetter-exploration-with-parameter-noise\u002F)]\n- Deep Reinforcement Learning from human preferences [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.03741)] [[article](https:\u002F\u002Fblog.openai.com\u002Fdeep-reinforcement-learning-from-human-preferences\u002F)]\n- Multi-Agent Actor-Critic for Mixed Cooperative-Competitive Environments [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.02275)] [[code](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fmultiagent-particle-envs)]\n- Self-Normalizing Neural Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.02515)] [[code](https:\u002F\u002Fgithub.com\u002Fbioinf-jku\u002FSNNs)]\n- Accurate, Large Minibatch SGD: Training ImageNet in 1 Hour [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.02677)]\n- A simple neural network module for relational reasoning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.01427)] [[article](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fneural-approach-relational-reasoning\u002F)]\n- Visual Interaction Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.01433)] [[article](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fneural-approach-relational-reasoning\u002F)]\n\n#### 2017-05\n\n- Supervised Learning of Universal Sentence Representations from Natural Language Inference Data [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.02364)]  [[code](https:\u002F\u002Fgithub.com\u002Ffacebookresearch\u002FInferSent)]\n- pix2code: Generating Code from a Graphical User Interface Screenshot [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.07962)] [[article](https:\u002F\u002Fuizard.io\u002Fresearch#pix2code)] [[code](https:\u002F\u002Fgithub.com\u002Ftonybeltramelli\u002Fpix2code)]\n- The Cramer Distance as a Solution to Biased Wasserstein Gradients [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.10743)]\n- Reinforcement Learning with a Corrupted Reward Channel [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.08417)]\n- Dilated Residual Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.09914)] [[code](https:\u002F\u002Fgithub.com\u002Ffyu\u002Fdrn)]\n- Bayesian GAN [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.09558)] [[code](https:\u002F\u002Fgithub.com\u002Fandrewgordonwilson\u002Fbayesgan\u002F)]\n- Gradient Descent Can Take Exponential Time to Escape Saddle Points [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.10412)] [[article](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2017\u002F08\u002F31\u002Fsaddle-efficiency\u002F)]\n- Thinking Fast and Slow with Deep Learning and Tree Search [[arXiv]()]\n- ParlAI: A Dialog Research Software Platform [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.06476)] [[code](https:\u002F\u002Fgithub.com\u002Ffacebookresearch\u002FParlAI)] [[article](https:\u002F\u002Fcode.facebook.com\u002Fposts\u002F266433647155520\u002Fparlai-a-new-software-platform-for-dialog-research\u002F)]\n- Semantically Decomposing the Latent Spaces of Generative Adversarial Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.07904)] [[article](https:\u002F\u002Faws.amazon.com\u002Fblogs\u002Fai\u002Fcombining-deep-learning-networks-gan-and-siamese-to-generate-high-quality-life-like-images\u002F)]\n- Look, Listen and Learn [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.08168)]\n- Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.07750)] [[code](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Fkinetics-i3d)]\n- Convolutional Sequence to Sequence Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.03122)] [[code](https:\u002F\u002Fgithub.com\u002Ffacebookresearch\u002Ffairseq)] [[code2](https:\u002F\u002Fgithub.com\u002Ffacebookresearch\u002Ffairseq-py)] [[article](https:\u002F\u002Fcode.facebook.com\u002Fposts\u002F1978007565818999\u002Fa-novel-approach-to-neural-machine-translation\u002F)]\n- The Kinetics Human Action Video Dataset [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.06950)] [[article](https:\u002F\u002Fdeepmind.com\u002Fresearch\u002Fopen-source\u002Fopen-source-datasets\u002Fkinetics\u002F)]\n- Safe and Nested Subgame Solving for Imperfect-Information Games [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.02955)]\n- Discrete Sequential Prediction of Continuous Actions for Deep RL [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.05035)]\n- Metacontrol for Adaptive Imagination-Based Optimization [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.02670)]\n- Efficient Parallel Methods for Deep Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.04862)]\n- Real-Time Adaptive Image Compression [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.05823)]\n\n#### 2017-04\n\n- General Video Game AI: Learning from Screen Capture [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.06945)]\n- Learning to Skim Text [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.06877)]\n- Get To The Point: Summarization with Pointer-Generator Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.04368)] [[code](https:\u002F\u002Fgithub.com\u002Fabisee\u002Fpointer-generator)] [[article](http:\u002F\u002Fwww.abigailsee.com\u002F2017\u002F04\u002F16\u002Ftaming-rnns-for-better-summarization.html)]\n- Adversarial Neural Machine Translation [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.06933)]\n- [Deep Q-learning from Demonstrations](notes\u002Fdqn-demonstrations.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.03732)]\n- Learning from Demonstrations for Real World Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.03732)]\n- DSLR-Quality Photos on Mobile Devices with Deep Convolutional Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.02470)] [[article](http:\u002F\u002Fpeople.ee.ethz.ch\u002F~ihnatova\u002F)] [[code](https:\u002F\u002Fgithub.com\u002Faiff22\u002FDPED)]\n- A Neural Representation of Sketch Drawings [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.03477)] [[code](https:\u002F\u002Fgithub.com\u002Ftensorflow\u002Fmagenta\u002Ftree\u002Fmaster\u002Fmagenta\u002Fmodels\u002Fsketch_rnn)] [[article](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F04\u002Fteaching-machines-to-draw.html)]\n- Automated Curriculum Learning for Neural Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.03003)]\n- Hierarchical Surface Prediction for 3D Object Reconstruction [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.00710)] [[article](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2017\u002F08\u002F23\u002Fhigh-quality-3d-obj-reconstruction\u002F)]\n- Neural Message Passing for Quantum Chemistry [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.01212)]\n- Learning to Generate Reviews and Discovering Sentiment [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.01444)] [[code](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fgenerating-reviews-discovering-sentiment)]\n- Best Practices for Applying Deep Learning to Novel Applications [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.01568)]\n\n#### 2017-03\n\n- Improved Training of Wasserstein GANs [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.00028)]\n- Evolution Strategies as a Scalable Alternative to Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.03864)]\n- Controllable Text Generation [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.00955)]\n- Neural Episodic Control [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.01988)]\n- [A Structured Self-attentive Sentence Embedding](notes\u002Fself_attention_embedding.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.03130)]\n- Multi-step Reinforcement Learning: A Unifying Algorithm [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.01327)]\n- Deep learning with convolutional neural networks for brain mapping and decoding of movement-related information from the human EEG [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.05051)]\n- FaSTrack: a Modular Framework for Fast and Guaranteed Safe Motion Planning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.07373)] [[article](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2017\u002F12\u002F05\u002Ffastrack\u002F)] [[article2](http:\u002F\u002Fsylviaherbert.com\u002Ffastrack\u002F)]\n- Massive Exploration of Neural Machine Translation Architectures [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.03906)] [[code](https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fseq2seq)]\n- Large Pose 3D Face Reconstruction from a Single Image via Direct Volumetric CNN Regression [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.07834)] [[article](http:\u002F\u002Faaronsplace.co.uk\u002Fpapers\u002Fjackson2017recon\u002F)] [[code](https:\u002F\u002Fgithub.com\u002FAaronJackson\u002Fvrn)]\n- Minimax Regret Bounds for Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.05449)]\n- Sharp Minima Can Generalize For Deep Nets [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.04933)]\n- Parallel Multiscale Autoregressive Density Estimation [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.03664)]\n- Neural Machine Translation and Sequence-to-sequence Models: A Tutorial [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.01619)]\n- Large-Scale Evolution of Image Classifiers [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.01041)]\n- FeUdal Networks for Hierarchical Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.01161)]\n- Evolving Deep Neural Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.00548)]\n- How to Escape Saddle Points Efficiently [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.00887)] [[article](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2017\u002F08\u002F31\u002Fsaddle-efficiency\u002F)]\n- Opening the Black Box of Deep Neural Networks via Information [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.00810)] [[video](https:\u002F\u002Fyoutu.be\u002FbLqJHjXihK8)]\n- Understanding Synthetic Gradients and Decoupled Neural Interfaces [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.00522)]\n- Learning to Optimize Neural Nets [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.00441)] [[article](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2017\u002F09\u002F12\u002Flearning-to-optimize-with-rl\u002F)]\n\n\n#### 2017-02\n\n- The Shattered Gradients Problem: If resnets are the answer, then what is the question? [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1702.08591)]\n- Neural Map: Structured Memory for Deep Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1702.08360)]\n- Bridging the Gap Between Value and Policy Based Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1702.08892)]\n- Deep Voice: Real-time Neural Text-to-Speech [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1702.07825)]\n- Beating the World's Best at Super Smash Bros. with Deep Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1702.06230)]\n- The Game Imitation: Deep Supervised Convolutional Networks for Quick Video Game AI [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1702.05663)]\n- Learning to Parse and Translate Improves Neural Machine Translation [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1702.03525)]\n- All-but-the-Top: Simple and Effective Postprocessing for Word Representations [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1702.01417)]\n- Deep Learning with Dynamic Computation Graphs [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1702.02181)]\n- Skip Connections as Effective Symmetry-Breaking [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1701.09175)]\n- odelSemi-Supervised QA with Generative Domain-Adaptive Nets [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1702.02206)]\n\n#### 2017-01\n\n- Wasserstein GAN [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1701.07875)]\n- Deep Reinforcement Learning: An Overview [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1701.07274)]\n- DyNet: The Dynamic Neural Network Toolkit [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1701.03980)]\n- DeepStack: Expert-Level Artificial Intelligence in No-Limit Poker [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1701.01724)]\n- NIPS 2016 Tutorial: Generative Adversarial Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1701.00160)]\n\n#### 2016-12\n\n- [A recurrent neural network without Chaos](notes\u002Frnn_no_chaos.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1612.06212)]\n- Language Modeling with Gated Convolutional Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1612.08083)]\n- EnhanceNet: Single Image Super-Resolution Through Automated Texture Synthesis [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1612.07919)] [[article](http:\u002F\u002Fwebdav.tuebingen.mpg.de\u002Fpixel\u002Fenhancenet\u002F)]\n- Learning from Simulated and Unsupervised Images through Adversarial Training [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1612.07828)]\n- How Grammatical is Character-level Neural Machine Translation? Assessing MT Quality with Contrastive Translation Pairs [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1612.04629)]\n- Improving Neural Language Models with a Continuous Cache [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1612.04426)]\n- DeepMind Lab [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1612.03801)] [[code](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Flab)]\n- Deep Learning of Robotic Tasks without a Simulator using Strong and Weak Human Supervision [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1612.01086)]\n- Knowing When to Look: Adaptive Attention via A Visual Sentinel for Image Captioning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1612.01887)]\n- Overcoming catastrophic forgetting in neural networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1612.00796)]\n\n#### 2016-11 (ICLR Edition)\n\n- Image-to-Image Translation with Conditional Adversarial Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.07004)]\n- [Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer](notes\u002Fmixture-experts.md) [[OpenReview](https:\u002F\u002Fopenreview.net\u002Fforum?id=B1ckMDqlg)]\n- Learning to reinforcement learn [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.05763)]\n- A Way out of the Odyssey: Analyzing and Combining Recent Insights for LSTMs [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.05104)]\n- [Adversarial Training Methods for Semi-Supervised Text Classification](notes\u002Fadversarial-text-classification.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1605.07725)]\n- Importance Sampling with Unequal Support [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.03451)]\n- Quasi-Recurrent Neural Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.01576)]\n- Capacity and Learnability in Recurrent Neural Networks [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=BydARw9ex)]\n- Unrolled Generative Adversarial Networks [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=BydrOIcle)]\n- Deep Information Propagation [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=H1W1UN9gg)]\n- Structured Attention Networks [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=HkE0Nvqlg)]\n- Incremental Sequence Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.03068)]\n- Delving into Transferable Adversarial Examples and Black-box Attacks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.02770)] [[code](https:\u002F\u002Fgithub.com\u002FReDeiPirati\u002Ftransferability-advdnn-pub)]\n- b-GAN: Unified Framework of Generative Adversarial Networks [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=S1JG13oee)]\n- A Joint Many-Task Model: Growing a Neural Network for Multiple NLP Tasks [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=SJZAb5cel)]\n- Categorical Reparameterization with Gumbel-Softmax [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.01144)]\n- Lip Reading Sentences in the Wild [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.05358)]\n\nReinforcement Learning:\n\n-Learning to reinforcement learn [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.05763)]\n- A Connection between Generative Adversarial Networks, Inverse Reinforcement Learning, and Energy-Based Models [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.03852)]\n- The Predictron: End-To-End Learning and Planning [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=BkJsCIcgl)]\n- [Third-Person Imitation Learning](notes\u002Fthird-person-imitation-learning.md) [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=B16dGcqlx)]\n- Generalizing Skills with Semi-Supervised Reinforcement Learning [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=ryHlUtqge)]\n- Sample Efficient Actor-Critic with Experience Replay [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=HyM25Mqel)]\n- [Reinforcement Learning with Unsupervised Auxiliary Tasks](notes\u002Frl-auxiliary-tasks.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.05397)]\n- Neural Architecture Search with Reinforcement Learning [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=r1Ue8Hcxg)]\n- Towards Information-Seeking Agents [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=SyW2QSige)]\n- Multi-Agent Cooperation and the Emergence of (Natural) Language [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=Hk8N3Sclg)]\n- Improving Policy Gradient by Exploring Under-appreciated Rewards [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=ryT4pvqll)]\n- Stochastic Neural Networks for Hierarchical Reinforcement Learning [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=B1oK8aoxe)]\n- Tuning Recurrent Neural Networks with Reinforcement Learning [[OpenReview](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.02796)]\n- RL^2: Fast Reinforcement Learning via Slow Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.02779)]\n- Learning Invariant Feature Spaces to Transfer Skills with Reinforcement Learning [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=Hyq4yhile)]\n- Learning to Perform Physics Experiments via Deep Reinforcement Learning [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=r1nTpv9eg)]\n- Reinforcement Learning through Asynchronous Advantage Actor-Critic on a GPU [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=r1VGvBcxl)]\n- Learning to Compose Words into Sentences with Reinforcement Learning[[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=Skvgqgqxe)]\n- Deep Reinforcement Learning for Accelerating the Convergence Rate [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=Syg_lYixe)]\n- [#Exploration: A Study of Count-Based Exploration for Deep Reinforcement Learning](notes\u002Fcount-based-exploration.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.04717)]\n- Learning to Compose Words into Sentences with Reinforcement Learning [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=Skvgqgqxe)]\n- Learning to Navigate in Complex Environments [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.03673)]\n- Unsupervised Perceptual Rewards for Imitation Learning [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=Bkul3t9ee)]\n- Q-Prop: Sample-Efficient Policy Gradient with An Off-Policy Critic [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=SJ3rcZcxl)]\n\n\nMachine Translation & Dialog\n\n- [Google's Multilingual Neural Machine Translation System: Enabling Zero-Shot Translation](notes\u002Fgnmt-multilingual.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.04558)]\n- [Neural Machine Translation with Reconstruction](notes\u002Fnmt-with-reconstruction.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.01874v1)]\n- Iterative Refinement for Machine Translation [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=r1y1aawlg)]\n- A Convolutional Encoder Model for Neural Machine Translation [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.02344)]\n- Improving Neural Language Models with a Continuous Cache [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=B184E5qee)]\n- Vocabulary Selection Strategies for Neural Machine Translation [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=Bk8N0RLxx)]\n- Towards an automatic Turing test: Learning to evaluate dialogue responses [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=HJ5PIaseg)]\n- Dialogue Learning With Human-in-the-Loop [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=HJgXCV9xx)]\n- Batch Policy Gradient Methods for Improving Neural Conversation Models [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=rJfMusFll)]\n- Learning through Dialogue Interactions [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=rkE8pVcle)]\n- [Dual Learning for Machine Translation](notes\u002Fdual-learning-mt.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.00179)]\n- Unsupervised Pretraining for Sequence to Sequence Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.02683)]\n\n\n\n#### 2016-10\n\n- Hybrid computing using a neural network with dynamic external memory [[nature](https:\u002F\u002Fwww.nature.com\u002Farticles\u002Fnature20101.epdf?author_access_token=ImTXBI8aWbYxYQ51Plys8NRgN0jAjWel9jnR3ZoTv0MggmpDmwljGswxVdeocYSurJ3hxupzWuRNeGvvXnoO8o4jTJcnAyhGuZzXJ1GEaD-Z7E6X_a9R-xqJ9TfJWBqz)] [[code](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Fdnc)]\n- Quantum Machine Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.09347)]\n- Understanding deep learning requires rethinking generalization [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.03530)]\n- Universal adversarial perturbations [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.08401)] [[code](https:\u002F\u002Fgithub.com\u002FLTS4\u002Funiversal)]\n- [Neural Machine Translation in Linear Time](notes\u002Fnmt-linear-time.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.10099)] [[code](https:\u002F\u002Fgithub.com\u002Ftensorflow\u002Ftensor2tensor)]\n- [Professor Forcing: A New Algorithm for Training Recurrent Networks](notes\u002Fprofessor-forcing.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.09038)]\n- Learning to Protect Communications with Adversarial Neural Cryptography [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.06918v1)]\n- Can Active Memory Replace Attention? [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.08613)]\n- [Using Fast Weights to Attend to the Recent Past](notes\u002Ffast-weight-to-attend.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.06258)]\n- [Fully Character-Level Neural Machine Translation without Explicit Segmentation](notes\u002Fconv-char-level-nmt.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.03017)]\n- [Diverse Beam Search: Decoding Diverse Solutions from Neural Sequence Models](notes\u002Fdiverse-beam-search.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.02424)]\n- Video Pixel Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.00527)]\n- Connecting Generative Adversarial Networks and Actor-Critic Methods [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.01945)]\n- [Learning to Translate in Real-time with Neural Machine Translation](notes\u002Flearning-to-translate-real-time.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.00388)]\n- Xception: Deep Learning with Depthwise Separable Convolutions [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.02357)]\n- Collective Robot Reinforcement Learning with Distributed Asynchronous Guided Policy Search [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.00673)]\n- [Pointer Sentinel Mixture Models](notes\u002Fpointer-sentinel-mixture.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1609.07843)]\n\n#### 2016-09\n\n- Towards Deep Symbolic Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1609.05518)]\n- HyperNetworks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1609.09106)]\n- Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1609.08144)]\n- Safe and Efficient Off-Policy Reinforcement Learning [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.02647)]\n- Playing FPS Games with Deep Reinforcement Learning [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1609.05521)]\n- [SeqGAN: Sequence Generative Adversarial Nets with Policy Gradient](notes\u002Fseq-gan.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1609.05473)]\n- Episodic Exploration for Deep Deterministic Policies: An Application to StarCraft Micromanagement Tasks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1609.02993)]\n- Energy-based Generative Adversarial Network [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1609.03126)]\n- Stealing Machine Learning Models via Prediction APIs [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1609.02943)]\n- Semi-Supervised Classification with Graph Convolutional Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1609.02907)]\n- WaveNet: A Generative Model For Raw Audio [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1609.03499)]\n- [Hierarchical Multiscale Recurrent Neural Networks](notes\u002Fhm-rnn.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1609.01704)]\n- End-to-End Reinforcement Learning of Dialogue Agents for Information Access [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1609.00777)]\n- Deep Neural Networks for YouTube Recommendations [[paper](https:\u002F\u002Fresearch.google.com\u002Fpubs\u002Fpub45530.html)]\n\n#### 2016-08\n\n- Semantics derived automatically from language corpora contain human-like biases [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1608.07187)]\n- Why does deep and cheap learning work so well? [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1608.08225)]\n- Machine Comprehension Using Match-LSTM and Answer Pointer [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1608.07905)]\n- Stacked Approximated Regression Machine: A Simple Deep Learning Approach [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1608.04062)]\n- Decoupled Neural Interfaces using Synthetic Gradients [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1608.05343)]\n- WikiReading: A Novel Large-scale Language Understanding Task over Wikipedia [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1608.03542)]\n- Temporal Attention Model for Neural Machine Translation [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1608.02927)]\n- Residual Networks of Residual Networks: Multilevel Residual Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1608.02908)]\n- [Learning Online Alignments with Continuous Rewards Policy Gradient](notes\u002Fonline-alignments-pg.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1608.01281)]\n\n#### 2016-07\n\n- [An Actor-Critic Algorithm for Sequence Prediction](notes\u002Factor-critic-sequence.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1607.07086)]\n- Cognitive Science in the era of Artificial Intelligence: A roadmap for reverse-engineering the infant language-learner [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1607.08723v1)]\n- [Recurrent Neural Machine Translation](notes\u002Frecurrent-nmt.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1607.08725)]\n- MS-Celeb-1M: A Dataset and Benchmark for Large-Scale Face Recognition [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1607.08221)]\n- [Layer Normalization](notes\u002Flayer-norm.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1607.06450)]\n- [Neural Machine Translation with Recurrent Attention Modeling](notes\u002Fnmt-rec-attention.md)  [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1607.05108)]\n- Neural Semantic Encoders [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1607.04315)]\n- [Attention-over-Attention Neural Networks for Reading Comprehension](notes\u002Fatt-over-att.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1607.04423)]\n- sk_p: a neural program corrector for MOOCs [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1607.02902)]\n- Recurrent Highway Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1607.03474)]\n- Bag of Tricks for Efficient Text Classification [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1607.01759)]\n- Context-Dependent Word Representation for Neural Machine Translation [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1607.00578)]\n- Dynamic Neural Turing Machine with Soft and Hard Addressing Schemes [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1607.00036)]\n\n#### 2016-06\n\n- Sequence-to-Sequence Learning as Beam-Search Optimization [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.02960)]\n- [Sequence-Level Knowledge Distillation](notes\u002Fseq-knowledge-distillation.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.07947)]\n- Policy Networks with Two-Stage Training for Dialogue Systems [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.03152)]\n- Towards an integration of deep learning and neuroscience [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.03813)]\n- On Multiplicative Integration with Recurrent Neural Networks [[arxiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.06630)]\n- [Wide & Deep Learning for Recommender Systems](wide-and-deep.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.07792)]\n- Online and Offline Handwritten Chinese Character Recognition [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.05763)]\n- Tutorial on Variational Autoencoders [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.05908)]\n- Concrete Problems in AI Safety [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.06565)]\n- Deep Reinforcement Learning Discovers Internal Models [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.05174v1)]\n- [SQuAD: 100,000+ Questions for Machine Comprehension of Text](notes\u002Fsquad.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.05250)]\n- Conditional Image Generation with PixelCNN Decoders [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.05328)]\n- Model-Free Episodic Control [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.04460)]\n- [Progressive Neural Networks](notes\u002Fprogressive-nn.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.04671)]\n- Improved Techniques for Training GANs [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.03498)] [[code](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fimproved-gan)]\n- Memory-Efficient Backpropagation Through Time [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.03401)]\n- InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.03657)]\n- Zero-Resource Translation with Multi-Lingual Neural Machine Translation [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.04164)]\n- Key-Value Memory Networks for Directly Reading Documents [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.03126)]\n- Deep Recurrent Models with Fast-Forward Connections for Neural Machine Translatin [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.04199)]\n- Learning to learn by gradient descent by gradient descent [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.04474)]\n- Learning Language Games through Interaction [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.02447)]\n- Zoneout: Regularizing RNNs by Randomly Preserving Hidden Activations [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.01305)]\n- Smart Reply: Automated Response Suggestion for Email [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.04870)]\n- Virtual Adversarial Training for Semi-Supervised Text Classification [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1605.07725)]\n- Deep Reinforcement Learning for Dialogue Generation [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.01541)]\n- Very Deep Convolutional Networks for Natural Language Processing [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.01781)]\n- Neural Net Models for Open-Domain Discourse Coherence [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.01545)]\n- Neural Architectures for Fine-grained Entity Type Classification [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.01341)]\n- Matching Networks for One Shot Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.04080)]\n- Cooperative Inverse Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.03137)] [[article](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2017\u002F08\u002F17\u002Fcooperatively-learning-human-values\u002F)]\n- Gated-Attention Readers for Text Comprehension [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.01549)]\n- [End-to-end LSTM-based dialog control optimized with supervised and reinforcement learning](notes\u002Fe2e-dialog-control-sl-rl.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.01269)]\n- Iterative Alternating Neural Attention for Machine Reading [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.02245)]\n- Memory-enhanced Decoder for Neural Machine Translation [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.02003)]\n- Multiresolution Recurrent Neural Networks: An Application to Dialogue Response Generation [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.00776)]\n- Learning to Optimize [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.01885)] [[article](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2017\u002F09\u002F12\u002Flearning-to-optimize-with-rl\u002F)]\n- [Natural Language Comprehension with the EpiReader](notes\u002Fepireader.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.02270)]\n- Conversational Contextual Cues: The Case of Personalization and History for Response Ranking [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.00372)]\n- Adversarially Learned Inference [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.00704)]\n- OpenAI Gym [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.01540)] [[code](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Flab)]\n- Neural Network Translation Models for Grammatical Error Correction [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.00189)]\n\n#### 2016-05\n\n- Hierarchical Memory Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1605.07427)]\n- Deep API Learning [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.08535)]\n- Wide Residual Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.07146)]\n- TensorFlow: A system for large-scale machine learning [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.08695)]\n- Learning Natural Language Inference using Bidirectional LSTM model and Inner-Attention [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.09090)]\n- Aspect Level Sentiment Classification with Deep Memory Network [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.08900)]\n- FractalNet: Ultra-Deep Neural Networks without Residuals [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1605.07648)]\n- Learning End-to-End Goal-Oriented Dialog [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.07683)]\n- One-shot Learning with Memory-Augmented Neural Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.06065)]\n- Deep Learning without Poor Local Minima [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.07110)]\n- AVEC 2016 - Depression, Mood, and Emotion Recognition Workshop and Challenge [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1605.01600)]\n- Data Programming: Creating Large Training Sets, Quickly [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.07723)]\n- Deeply-Fused Nets [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.07716)]\n- Deep Portfolio Theory [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.07230)]\n- Unsupervised Learning for Physical Interaction through Video Prediction [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.07157)]\n- Movie Description [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.03705)]\n\n\n#### 2016-04\n\n- Higher Order Recurrent Neural Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1605.00064)]\n- Joint Line Segmentation and Transcription for End-to-End Handwritten Paragraph Recognition [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1604.08352)]\n- Hierarchical Deep Reinforcement Learning: Integrating Temporal Abstraction and Intrinsic Motivation [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1604.06057)]\n- The IBM 2016 English Conversational Telephone Speech Recognition System [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1604.08242)]\n- Dialog-based Language Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1604.06045)]\n- Multilingual Part-of-Speech Tagging with Bidirectional Long Short-Term Memory Models and Auxiliary Loss [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1604.05529)]\n- Sentence-Level Grammatical Error Identification as Sequence-to-Sequence Correction [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1604.04677)]\n- A Network-based End-to-End Trainable Task-oriented Dialogue System [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1604.04562)]\n- Visual Storytelling [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1604.03968)]\n- Improving the Robustness of Deep Neural Networks via Stability Training [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1604.04326)]\n- [Bridging the Gaps Between Residual Learning, Recurrent Neural Networks and Visual Cortex](notes\u002Fbridging-gap-resnet-rnn.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1604.03640)]\n- Scan, Attend and Read: End-to-End Handwritten Paragraph Recognition with MDLSTM Attention [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1604.03286)]\n- [Sentence Level Recurrent Topic Model: Letting Topics Speak for Themselves](notes\u002Fslrtm.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1604.02038)]\n- [Achieving Open Vocabulary Neural Machine Translation with Hybrid Word-Character Models](notes\u002Fopen-vocab-nmt-hybrid-word-character.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1604.00788)]\n- [Building Machines That Learn and Think Like People](notes\u002Fbuilding-machines-that-learn-and-think-like-people.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1604.00289)]\n- A Semisupervised Approach for Language Identification based on Ladder Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1604.00317)]\n- [Deep Networks with Stochastic Depth](notes\u002Fstochastic-depth.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.09382)]\n- PHOCNet: A Deep Convolutional Neural Network for Word Spotting in Handwritten Documents [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1604.00187)]\n\n\n#### 2016-03\n\n- Improving Information Extraction by Acquiring External Evidence with Reinforcement Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1603.07954)]\n- A Fast Unified Model for Parsing and Sentence Understanding [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.06021)]\n- [Latent Predictor Networks for Code Generation](notes\u002Flatent-predictor-networks.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.06744)]\n- Attend, Infer, Repeat: Fast Scene Understanding with Generative Models [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.08575)]\n- Recurrent Batch Normalization [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.09025)]\n- Neural Language Correction with Character-Based Attention [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.09727)]\n- [Incorporating Copying Mechanism in Sequence-to-Sequence Learning](notes\u002Fcopynet.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.06393)]\n- How NOT To Evaluate Your Dialogue System [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.08023)]\n- [Adaptive Computation Time for Recurrent Neural Networks](notes\u002Fact-rnn.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.08983)]\n- A guide to convolution arithmetic for deep learning [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.07285)]\n- Colorful Image Colorization [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.08983)]\n- Unsupervised Learning of Visual Representations by Solving Jigsaw Puzzles [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.09246)]\n- Generating Factoid Questions With Recurrent Neural Networks: The 30M Factoid Question-Answer Corpus [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.06807)]\n- A Persona-Based Neural Conversation Model [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.06155)]\n- [A Character-level Decoder without Explicit Segmentation for Neural Machine Translation](notes\u002Fchar-level-decoder.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.06147)]\n- Multi-Task Cross-Lingual Sequence Tagging from Scratch [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.06270)]\n- Neural Variational Inference for Text Processing [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06038)]\n- Recurrent Dropout without Memory Loss [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.05118)]\n- One-Shot Generalization in Deep Generative Models [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.05106)]\n- Recursive Recurrent Nets with Attention Modeling for OCR in the Wild [[arXiv](Recursive Recurrent Nets with Attention Modeling for OCR in the Wild)]\n- A New Method to Visualize Deep Neural Networks [[arXiv](A New Method to Visualize Deep Neural Networks)]\n- Neural Architectures for Named Entity Recognition [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.01360)]\n- End-to-end Sequence Labeling via Bi-directional LSTM-CNNs-CRF [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.01354)]\n- Character-based Neural Machine Translation [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.00810)]\n- Learning Word Segmentation Representations to Improve Named Entity Recognition for Chinese Social Media [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.00786)]\n\n#### 2016-02\n\n- Architectural Complexity Measures of Recurrent Neural Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.08210)]\n- Weight Normalization: A Simple Reparameterization to Accelerate Training of Deep Neural Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.07868)]\n- Recurrent Neural Network Grammars [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.07776)]\n- Visual Genome: Connecting Language and Vision Using Crowdsourced Dense Image Annotations [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.07332)]\n- [Contextual LSTM (CLSTM) models for Large scale NLP tasks](notes\u002Fclstm-large-scale.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.06291)]\n- Sequence-to-Sequence RNNs for Text Summarization [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.06023)]\n- Extraction of Salient Sentences from Labelled Documents [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1412.6815)]\n- Learning Distributed Representations of Sentences from Unlabelled Data [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.03483)]\n- Benefits of depth in neural networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.04485)]\n- [Associative Long Short-Term Memory](notes\u002Fassociative-lstm.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.03032)]\n- Why Should I Trust You?\": Explaining the Predictions of Any Classifier [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1602.04938)] [[code](https:\u002F\u002Fgithub.com\u002Fmarcotcr\u002Flime)]\n- Generating images with recurrent adversarial networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.05110)]\n- [Exploring the Limits of Language Modeling](notes\u002Fexploring-the-limits-of-lm.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.02410)]\n- Swivel: Improving Embeddings by Noticing What’s Missing [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.02215)]\n- [WebNav: A New Large-Scale Task for Natural Language based Sequential Decision Making](notes\u002Fwebnav.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.02261)]\n- [Efficient Character-level Document Classification by Combining Convolution and Recurrent Layers](notes\u002Fefficient-char-level-document-classification-cnn-rnn.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.00367)]\n- Gradient Descent Converges to Minimizers [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1602.04915)] [[article](http:\u002F\u002Fwww.offconvex.org\u002F2016\u002F03\u002F24\u002Fsaddles-again\u002F)]\n- BinaryNet: Training Deep Neural Networks with Weights and Activations Constrained to +1 or -1 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.02830)]\n- Learning Discriminative Features via Label Consistent Neural Network [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.01168)]\n\n#### 2016-01\n\n- What’s your ML test score? A rubric for ML production systems [[Research at Google](https:\u002F\u002Fresearch.google.com\u002Fpubs\u002Fpub45742.html)]\n- Pixel Recurrent Neural Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.06759)]\n- Bitwise Neural Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.06071)]\n- Long Short-Term Memory-Networks for Machine Reading [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.06733)]\n- Coverage-based Neural Machine Translation [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.04811)]\n- Understanding Deep Convolutional Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.04920)]\n- Training Recurrent Neural Networks by Diffusion [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.04114)]\n- Automatic Description Generation from Images: A Survey of Models, Datasets, and Evaluation Measures [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.03896)]\n- [Multi-Way, Multilingual Neural Machine Translation with a Shared Attention Mechanism](notes\u002Fmulti-way-nmt-shared-attention.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.01073)]\n- [Recurrent Memory Network for Language Modeling](notes\u002Frmn-language-modeling.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.01272)]\n- Language to Logical Form with Neural Attention [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.01280)]\n- Learning to Compose Neural Networks for Question Answering [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.01705)]\n- The Inevitability of Probability: Probabilistic Inference in Generic Neural Networks Trained with Non-Probabilistic Feedback [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.03060)]\n- COCO-Text: Dataset and Benchmark for Text Detection and Recognition in Natural Images [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.07140)]\n- Survey on the attention based RNN model and its applications in computer vision [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.06823)]\n\n#### 2015-12\n\nNLP\n\n- [Strategies for Training Large Vocabulary Neural Language Models](notes\u002Fstrategies-for-training-large-vocab-lm.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1512.04906)]\n- [Multilingual Language Processing From Bytes](notes\u002Fmultilingual-language-processing-from-bytes.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1512.00103)]\n- [Learning Document Embeddings by Predicting N-grams for Sentiment Classification of Long Movie Reviews](notes\u002Flearning-document-embeddings-ngrams.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1512.08183)]\n- [Target-Dependent Sentiment Classification with Long Short Term Memory](notes\u002Ftarget-dependent-sentiment-lstm.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1512.01100)]\n- Reading Text in the Wild with Convolutional Neural Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1412.1842)]\n\nVision\n\n- [Deep Residual Learning for Image Recognition](notes\u002Fdeep-residual-learning.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1512.03385)]\n- Rethinking the Inception Architecture for Computer Vision [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1512.00567)]\n- Inside-Outside Net: Detecting Objects in Context with Skip Pooling and Recurrent Neural Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1512.04143)]\n- Deep Speech 2: End-to-End Speech Recognition in English and Mandarin [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1512.02595)]\n\n\n#### 2015-11\n\nNLP\n\n- [Deep Reinforcement Learning with a Natural Language Action Space](notes\u002Fdrl-nlp-action.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1511.04636)]\n- Sequence Level Training with Recurrent Neural Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06732)]\n- [Teaching Machines to Read and Comprehend](notes\u002Fteaching-machines-to-read-and-comprehend.md) [[arxiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1506.03340)]\n- [Semi-supervised Sequence Learning](notes\u002Fsemi-supervised-sequence-learning.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.01432)]\n- [Multi-task Sequence to Sequence Learning](notes\u002Fmultitask-seq2seq.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06114)]\n- [Alternative structures for character-level RNNs](notes\u002Falternative-structure-char-rnn.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06303)]\n- [Larger-Context Language Modeling](notes\u002Flarger-context-lm.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.03729)]\n- [A Unified Tagging Solution: Bidirectional LSTM Recurrent Neural Network with Word Embedding](notes\u002Funified-tagging-blstm.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.00215)]\n- Towards Universal Paraphrastic Sentence Embeddings [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.08198)]\n- BlackOut: Speeding up Recurrent Neural Network Language Models With Very Large Vocabularies [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06909)]\n- Sequence Level Training with Recurrent Neural Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06732)]\n- Natural Language Understanding with Distributed Representation [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.07916)]\n- sense2vec - A Fast and Accurate Method for Word Sense Disambiguation In Neural Word Embeddings [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06388)]\n- LSTM-based Deep Learning Models for non-factoid answer selection [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.04108)]\n\nPrograms\n\n- Neural Random-Access Machines [[arxiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06392)]\n- Neural Programmer: Inducing Latent Programs with Gradient Descent [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.04834)]\n- Neural Programmer-Interpreters [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06279)]\n- Learning Simple Algorithms from Examples [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.07275)]\n- Neural GPUs Learn Algorithms [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.08228)] [[code](https:\u002F\u002Fgithub.com\u002Ftensorflow\u002Ftensor2tensor)]\n- On Learning to Think: Algorithmic Information Theory for Novel Combinations of Reinforcement Learning Controllers and Recurrent Neural World Models [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.09249)]\n\nVision\n\n- ReSeg: A Recurrent Neural Network for Object Segmentation [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.07053)]\n- Deconstructing the Ladder Network Architecture [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06430)]\n- Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06434)]\n- Multi-Scale Context Aggregation by Dilated Convolutions [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1511.07122)] [[code](https:\u002F\u002Fgithub.com\u002Ffyu\u002Fdrn)]\n\nGeneral\n\n- Towards Principled Unsupervised Learning [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06440)]\n- Dynamic Capacity Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.07838)]\n- [Generating Sentences from a `ous Space](notes\u002Fgenerating-sentences-cont-space.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06349)]\n- Net2Net: Accelerating Learning via Knowledge Transfer [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.05641)]\n- A Roadmap towards Machine Intelligence [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.08130)]\n- Session-based Recommendations with Recurrent Neural Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06939)]\n- Regularizing RNNs by Stabilizing Activations [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.08400)]\n\n\n#### 2015-10\n\n- [A Sensitivity Analysis of (and Practitioners' Guide to) Convolutional Neural Networks for Sentence Classification](notes\u002Fsensitivity-analysis-cnn-sentence-classification.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1510.03820)]\n- [Attention with Intention for a Neural Network Conversation Model](notes\u002Fattention-with-intention.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1510.08565)]\n- Part-of-Speech Tagging with Bidirectional Long Short-Term Memory Recurrent Neural Network [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1510.06168)]\n- A Survey: Time Travel in Deep Learning Space: An Introduction to Deep Learning Models and How Deep Learning Models Evolved from the Initial Ideas [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1510.04781)]\n- A Primer on Neural Network Models for Natural Language Processing [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1510.00726)]\n- [A Diversity-Promoting Objective Function for Neural Conversation Models](notes\u002Fdiversity-promoting-objective-ncm.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1510.03055)]\n\n\n#### 2015-09\n\n- [Character-level Convolutional Networks for Text Classification](notes\u002Fcharacter-level-cnn-for-text-classification.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1509.01626)]\n- [A Neural Attention Model for Abstractive Sentence Summarization](notes\u002Fneural-attention-model-for-abstractive-sentence-summarization.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1509.00685)]\n- Poker-CNN: A Pattern Learning Strategy for Making Draws and Bets in Poker Games [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1509.06731)]\n\n#### 2015-08\n\n- [Neural Machine Translation of Rare Words with Subword Units](notes\u002Fnmt-subword.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1508.07909)] [[code](https:\u002F\u002Fgithub.com\u002Frsennrich\u002Fsubword-nmt)]\n- Listen, Attend and Spell [[arxiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1508.01211)]\n- [Character-Aware Neural Language Models](notes\u002Fcharacter-aware-nlm.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1508.06615)]\n- Improved Transition-Based Parsing by Modeling Characters instead of Words with LSTMs [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1508.00657)]\n- Finding Function in Form: Compositional Character Models for Open Vocabulary Word Representation [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1508.02096)]\n- [Effective Approaches to Attention-based Neural Machine Translation](notes\u002Feffective-approaches-nmt-attention.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1508.04025)]\n\n#### 2015-07\n\n- [Building End-To-End Dialogue Systems Using Generative Hierarchical Neural Network Models](e2e-dialog-ghnnm.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1507.04808)]\n- Semi-Supervised Learning with Ladder Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1507.02672)]\n- [Document Embedding with Paragraph Vectors](notes\u002Fdocument-embedding-with-pv.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1507.07998)]\n- [Training Very Deep Networks](notes\u002Ftraining-very-deep-networks.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1507.06228)]\n\n#### 2015-06\n\n- Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1506.02142)]\n- [A Neural Network Approach to Context-Sensitive Generation of Conversational Responses](notes\u002Fnn-context-sentitive-responses.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1506.06714)]\n- [Document Embedding with Paragraph Vectors](notes\u002Fdocument-embedding-with-pv.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1507.07998)]\n- [A Neural Conversational Model](notes\u002Fneural-conversational-model.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1506.05869)]\n- [Skip-Thought Vectors](notes\u002Fskip-thought-vectors.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1506.06726)]\n- [Pointer Networks](notes\u002Fpointer-networks.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1506.03134)]\n- [Spatial Transformer Networks](notes\u002Fspatial-transformer-networks.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1506.02025)]\n- Tree-structured composition in neural networks without tree-structured architectures [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1506.04834)]\n- Visualizing and Understanding Neural Models in NLP [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1506.01066)]\n- Learning to Transduce with Unbounded Memory [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1506.02516)]\n- Ask Me Anything: Dynamic Memory Networks for Natural Language Processing [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1506.07285)]\n- [Deep Knowledge Tracing](notes\u002Fdeep-knowledge-tracing.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1506.05908)]\n\n#### 2015-05\n\n- [ReNet: A Recurrent Neural Network Based Alternative to Convolutional Networks](notes\u002Frenet-rnn-alternative-to-convnet.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1505.00393)]\n- Reinforcement Learning Neural Turing Machines [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1505.00521)]\n\n#### 2015-04\n\n- Correlational Neural Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1504.07225)]\n\n#### 2015-03\n\n\n- [Distilling the Knowledge in a Neural Network](notes\u002Fdistilling-the-knowledge-in-a-nn.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1503.02531)]\n- [End-To-End Memory Networks](notes\u002Fend-to-end-memory-networks.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1503.08895)]\n- [Neural Responding Machine for Short-Text Conversation](notes\u002Fneural-responding-machine.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1503.02364)]\n- [Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift](notes\u002Fbatch-normalization.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1502.03167)]\n- Escaping From Saddle Points --- Online Stochastic Gradient for Tensor Decomposition [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1503.02101)] [[article](Escaping from Saddle Points)]\n\n\n#### 2015-02\n\n- Human-level control through deep reinforcement\nlearning [[Nature](https:\u002F\u002Fweb.stanford.edu\u002Fclass\u002Fpsych209\u002FReadings\u002FMnihEtAlHassibis15NatureControlDeepRL.pdf)] [[code](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Fdqn)]\n- [Text Understanding from Scratch](notes\u002Ftext-understanding-from-scratch.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1502.01710)]\n- [Show, Attend and Tell: Neural Image Caption Generation with Visual Attention](notes\u002Fshow-attend-tell.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1502.03044)]\n\n#### 2015-01\n\n- Hidden Technical Debt in Machine Learning Systems [[NIPS](https:\u002F\u002Fpapers.nips.cc\u002Fpaper\u002F5656-hidden-technical-debt-in-machine-learning-systems.pdf)]\n\n#### 2014-12\n\n- Learning Longer Memory in Recurrent Neural Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1412.7753)]\n- [Neural Turing Machines](notes\u002Fneural-turing-machines.md) [[arxiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1410.5401)]\n- [Grammar as a Foreign Langauage](notes\u002Fgrammar-as-a-foreign-language.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1412.7449)]\n- [On Using Very Large Target Vocabulary for Neural Machine Translation](notes\u002Fon-using-very-large-target-vocabulary-for-nmt.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1412.2007)]\n- Effective Use of Word Order for Text Categorization with Convolutional Neural Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1412.1058v1)]\n- Multiple Object Recognition with Visual Attention [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1412.7755)]\n\n#### 2014-11\n\n- The Loss Surfaces of Multilayer Networks [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1412.0233)]\n\n#### 2014-10\n\n- [Learning to Execute](notes\u002Flearning-to-execute.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1410.4615)]\n\n#### 2014-09\n\n- [Sequence to Sequence Learning with Neural Networks](notes\u002Fseq2seq-with-neural-networks.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1409.3215)]\n- [Neural Machine Translation by Jointly Learning to Align and Translate](notes\u002Fnmt-jointly-learning-to-align-and-translate.md) [[arxiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1409.0473)]\n- [On the Properties of Neural Machine Translation: Encoder-Decoder Approaches](notes\u002Fproperties-of-neural-mt.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1409.1259)]\n- [Recurrent Neural Network Regularization](notes\u002Frnn-regularization.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1409.2329)]\n- Very Deep Convolutional Networks for Large-Scale Image Recognition [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1409.1556)]\n- Going Deeper with Convolutions [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1409.4842)]\n\n#### 2014-08\n\n- Convolutional Neural Networks for Sentence Classification [[arxiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1408.5882)]\n\n#### 2014-07\n\n#### 2014-06\n\n- [Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation](notes\u002Flearning-phrase-representations.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1406.1078)]\n- [Recurrent Models of Visual Attention](notes\u002Frecurrent-models-of-visual-attention.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1406.6247)]\n- Generative Adversarial Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1406.2661)]\n\n#### 2014-05\n\n- [Distributed Representations of Sentences and Documents](notes\u002Fdistributed-representations-of-sentences-and-documents.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1405.4053)]\n\n#### 2014-04\n\n- A Convolutional Neural Network for Modelling Sentences [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1404.2188)]\n\n#### 2014-03\n\n#### 2014-02\n\n#### 2014-01\n\n- Machine Learning: The High Interest Credit Card of Technical Debt [[Research at Google](https:\u002F\u002Fresearch.google.com\u002Fpubs\u002Fpub43146.html)]\n\n#### 2013\n\n- Visualizing and Understanding Convolutional Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1311.2901)]\n- DeViSE: A Deep Visual-Semantic Embedding Model [[pub](http:\u002F\u002Fresearch.google.com\u002Fpubs\u002Fpub41473.html)]\n- Maxout Networks [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1302.4389)]\n- Exploiting Similarities among Languages for Machine Translation [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1309.4168)]\n- Efficient Estimation of Word Representations in Vector Space [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1301.3781)]\n\n\n#### 2011\n\n- Natural Language Processing (almost) from Scratch [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1103.0398)]\n\n","#### 2018年2月\n\n- 深度学习所需的矩阵微积分 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1802.01528v2)]\n- 用于图像分类器架构搜索的正则化进化算法 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1802.01548)]\n- 在线学习：全面综述 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1802.02871)]\n- 深度学习的视觉可解释性：综述 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1802.00614)]\n- 行为即一切——以感觉运动协变性表征概念 [[论文](https:\u002F\u002Fwww.vicarious.com\u002Fwp-content\u002Fuploads\u002F2018\u002F01\u002FAAAI18-pixelworld.pdf)] [[文章](https:\u002F\u002Fwww.vicarious.com\u002F2018\u002F02\u002F07\u002Flearning-concepts-through-sensorimotor-interactions\u002F)] [[代码](https:\u002F\u002Fgithub.com\u002Fvicariousinc\u002Fpixelworld)]\n- IMPALA：基于重要性加权演员-学习者架构的可扩展分布式深度强化学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1802.01561)] [[文章](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fimpala-scalable-distributed-deeprl-dmlab-30\u002F)] [[代码](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Flab\u002Ftree\u002Fmaster\u002Fgame_scripts\u002Flevels\u002Fcontributed\u002Fdmlab30)]\n- DeepType：通过神经类型系统演化实现多语言实体链接 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1802.01021)] [[文章](https:\u002F\u002Fblog.openai.com\u002Fdiscovering-types-for-entity-disambiguation\u002F)] [[代码](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fdeeptype)]\n- DensePose：野外密集人体姿态估计 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1802.00434)] [[文章](http:\u002F\u002Fdensepose.org\u002F)]\n\n#### 2018年1月\n\n- 嵌套LSTM [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.10308)]\n- 通过总结长序列生成维基百科 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.10198)]\n- 面向电子健康记录的可扩展且精确的深度学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.07860)]\n- 基于条件协方差最小化的核特征选择 [[NIPS论文](https:\u002F\u002Fpapers.nips.cc\u002Fpaper\u002F7270-kernel-feature-selection-via-conditional-covariance-minimization.pdf)] [[文章](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2018\u002F01\u002F23\u002Fkernels\u002F)] [[代码](https:\u002F\u002Fgithub.com\u002FJianbo-Lab\u002FCCM)]\n- Psychlab：面向深度强化学习智能体的心理学实验室 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.08116)] [[文章](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fopen-sourcing-psychlab\u002F)] [[代码](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Flab\u002Ftree\u002Fmaster\u002Fgame_scripts\u002Flevels\u002Fcontributed\u002Fpsychlab)]\n- 针对文本分类的微调语言模型 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.06146)] [[代码]()]（即将发布）\n- 深度学习：应用数学家入门 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.05894v1)]\n- 先天性、AlphaZero与人工智能 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.05667)]\n- 计算机能创作艺术吗？[[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.04486)]\n- eCommerceGAN：用于电子商务的生成对抗网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.03244)]\n- 强化学习中的期望策略梯度 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.03326)]\n- DroNet：通过驾驶学习飞行 [[UZH文档](http:\u002F\u002Frpg.ifi.uzh.ch\u002Fdocs\u002FRAL18_Loquercio.pdf)] [[文章](http:\u002F\u002Frpg.ifi.uzh.ch\u002Fdronet.html)] [[代码](https:\u002F\u002Fgithub.com\u002Fuzh-rpg\u002Frpg_public_dronet)]\n- 非对称博弈的对称分解 [[Scientific Reports](https:\u002F\u002Fwww.nature.com\u002Farticles\u002Fs41598-018-19194-4)] [[文章](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fgame-theory-insights-asymmetric-multi-agent-games\u002F)]\n- 软演员-评论家：基于随机演员的离策略最大熵深度强化学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.01290)] [[代码](https:\u002F\u002Fgithub.com\u002Fhaarnoja\u002Fsac)]\n- SBNet：用于快速推理的稀疏块网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fpdf\u002F1801.02108.pdf)] [[文章](https:\u002F\u002Feng.uber.com\u002Fsbnet\u002F)] [[代码](https:\u002F\u002Fgithub.com\u002Fuber\u002Fsbnet)]\n- DeepMind控制套件 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.00690)] [[代码](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Fdm_control)]\n- 深度学习：批判性评析 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.00631)]\n\n\n#### 2017年12月\n\n- 对抗性补丁 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.09665)]\n- 只需CNN即可 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.09662)]\n- 从物理人机交互中学习机器人目标 [[论文](http:\u002F\u002Fproceedings.mlr.press\u002Fv78\u002Fbajcsy17a\u002Fbajcsy17a.pdf)] [[文章](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2018\u002F02\u002F06\u002Fphri\u002F)]\n- NarrativeQA阅读理解挑战 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.07040v1)] [[数据集](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Fnarrativeqa)]\n- 会发声的对象 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.06651)]\n- 基于梅尔频谱图预测的WaveNet自然TTS合成 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.05884)] [[文章](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F12\u002Ftacotron-2-generating-human-like-speech.html)] [[文章2](https:\u002F\u002Fgoogle.github.io\u002Ftacotron\u002Fpublications\u002Ftacotron2\u002Findex.html)]\n- 深度神经进化：遗传算法是训练深度神经网络进行强化学习的一种有竞争力的替代方案 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.06567)] [[文章](https:\u002F\u002Feng.uber.com\u002Fdeep-neuroevolution\u002F)] [[代码](https:\u002F\u002Fgithub.com\u002Fuber-common\u002Fdeep-neuroevolution)]\n- 通过一群追求新颖性的智能体改进深度强化学习进化策略中的探索 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.06560)] [[文章](https:\u002F\u002Feng.uber.com\u002Fdeep-neuroevolution\u002F)] [[代码](https:\u002F\u002Fgithub.com\u002Fuber-common\u002Fdeep-neuroevolution)]\n- 无限制德州扑克中超越人类水平的人工智能：Libratus击败顶尖职业选手 [[Science](http:\u002F\u002Fscience.sciencemag.org\u002Fcontent\u002Fearly\u002F2017\u002F12\u002F15\u002Fscience.aao1733)]\n- 深度学习的数学 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.04741)]\n- 基于序列到序列模型的最先进语音识别 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.01769)] [[文章](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F12\u002Fimproving-end-to-end-models-for-speech.html)]\n- Peephole：在训练前预测网络性能 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.03351)]\n- 审议网络：推动神经机器翻译的前沿 [[微软研究](https:\u002F\u002Fwww.microsoft.com\u002Fen-us\u002Fresearch\u002Fpublication\u002Fdeliberation-networks-sequence-generation-beyond-one-pass-decoding\u002F)] [[文章](https:\u002F\u002Fwww.microsoft.com\u002Fen-us\u002Fresearch\u002Fblog\u002Fdeliberation-networks\u002F)]\n- 用于块稀疏权重的GPU内核 [[OpenAI研究](https:\u002F\u002Fs3-us-west-2.amazonaws.com\u002Fopenai-assets\u002Fblocksparse\u002Fblocksparsepaper.pdf)] [[文章](https:\u002F\u002Fblog.openai.com\u002Fblock-sparse-gpu-kernels\u002F)] [[代码](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fblocksparse)]\n- 通过通用强化学习算法的自我博弈掌握国际象棋和将棋 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.01815)]\n- 深度学习的规模效应具有可预测性，经验证明 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1712.00409)] [[文章](http:\u002F\u002Fresearch.baidu.com\u002Fdeep-learning-scaling-predictable-empirically\u002F)]\n\n#### 2017年11月\n\n- 基于条件生成对抗网络的高分辨率图像合成与语义操控 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.11585)] [[文章](https:\u002F\u002Ftcwang0509.github.io\u002Fpix2pixHD\u002F)] [[代码](https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fpix2pixHD)]\n- StarGAN：用于多领域图像到图像转换的统一生成对抗网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.09020)] [[代码](https:\u002F\u002Fgithub.com\u002Fyunjey\u002FStarGAN\u002F)]\n- 神经网络的群体基础训练 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.09846)] [[文章](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fpopulation-based-training-neural-networks\u002F)]\n- 将神经网络蒸馏为软决策树 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.09784)]\n- 神经文本生成：实用指南 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.09534)]\n- 并行WaveNet：快速高保真语音合成 [[DeepMind文档](https:\u002F\u002Fdeepmind.com\u002Fdocuments\u002F131\u002FDistilling_WaveNet.pdf)] [[文章](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fhigh-fidelity-speech-synthesis-wavenet\u002F)]\n- CheXNet：基于深度学习的胸部X光片肺炎检测达到放射科医生水平 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.05225)] [[文章](https:\u002F\u002Fstanfordmlgroup.github.io\u002Fprojects\u002Fchexnet\u002F)]\n- 非局部神经网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.07971)]\n- 深度图像先验 [[论文](https:\u002F\u002Fsites.skoltech.ru\u002Fapp\u002Fdata\u002Fuploads\u002Fsites\u002F25\u002F2017\u002F11\u002Fdeep_image_prior.pdf)] [[文章](https:\u002F\u002Fdmitryulyanov.github.io\u002Fdeep_image_prior)] [[代码](https:\u002F\u002Fgithub.com\u002FDmitryUlyanov\u002Fdeep-image-prior)]\n- 在线深度学习：实时学习深度神经网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.03705)]\n- 从噪声数据中学习解释性规则 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.04574)]\n- 利用深度学习改善姑息治疗 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.06402)] [[文章](https:\u002F\u002Fstanfordmlgroup.github.io\u002Fprojects\u002Fimproving-palliative-care\u002F)]\n- VoxelNet：基于点云的三维目标检测端到端学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.06396)]\n- 用于机器翻译的加权Transformer网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.02132)] [[文章](https:\u002F\u002Feinstein.ai\u002Fresearch\u002Fweighted-transformer)]\n- 非自回归神经机器翻译 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.02281)] [[文章](https:\u002F\u002Feinstein.ai\u002Fresearch\u002Fnon-autoregressive-neural-machine-translation)]\n- 块稀疏循环神经网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.02782)]\n- 多智能体强化学习的统一博弈论方法 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.00832)]\n- 神经离散表征学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.00937)] [[文章](https:\u002F\u002Favdnoord.github.io\u002Fhomepage\u002Fvqvae\u002F)]\n- 不要衰减学习率，增大批量大小 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.00489)]\n- 用于高效架构搜索的层次化表示 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.00436)]\n\n#### 2017年10月\n\n- 仅使用单语语料库的无监督机器翻译 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1711.00043)]\n- 胶囊之间的动态路由 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.09829)]\n- 一种以高数据效率训练并破解基于文本的验证码的生成式视觉模型 [[Science](http:\u002F\u002Fscience.sciencemag.org\u002Fcontent\u002Fearly\u002F2017\u002F10\u002F26\u002Fscience.aag2612.full)] [[文章](https:\u002F\u002Fwww.vicarious.com\u002F2017\u002F10\u002F26\u002Fcommon-sense-cortex-and-captcha\u002F)] [[代码](https:\u002F\u002Fgithub.com\u002Fvicariousinc\u002Fscience_rcn)]\n- 理解具身语言学习智能体 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.09867)]\n- 计划，快与慢：自适应实时安全轨迹规划框架 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.04731)] [[文章](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2017\u002F12\u002F05\u002Ffastrack\u002F)] [[代码](https:\u002F\u002Fgithub.com\u002FHJReachability)]（即将发布）\n- 通过“吃掉”整个EXE文件进行恶意软件检测 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.09435)] [[文章](https:\u002F\u002Fdevblogs.nvidia.com\u002Fmalware-detection-neural-networks\u002F)]\n- 改进质量、稳定性和多样性的GAN渐进式增长方法 [[NVIDIA研究](http:\u002F\u002Fresearch.nvidia.com\u002Fsites\u002Fdefault\u002Ffiles\u002Fpubs\u002F2017-10_Progressive-Growing-of\u002F\u002Fkarras2017gan-paper.pdf)] [[文章](http:\u002F\u002Fresearch.nvidia.com\u002Fpublication\u002F2017-10_Progressive-Growing-of)] [[代码](https:\u002F\u002Fgithub.com\u002Ftkarras\u002Fprogressive_growing_of_gans)]\n- 共享层次结构的元学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.09767)] [[文章](https:\u002F\u002Fblog.openai.com\u002Flearning-a-hierarchy\u002F)] [[代码](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fmlsh)]\n- Deep Voice 3：2000位说话人的神经网络文本转语音系统 [[arXiv](http:\u002F\u002Fresearch.baidu.com\u002Fdeep-voice-3-2000-speaker-neural-text-speech\u002F)] [[文章](http:\u002F\u002Fresearch.baidu.com\u002Fdeep-voice-3-2000-speaker-neural-text-speech\u002F)]\n- AVA：一个时空局部化的原子视觉动作视频数据集 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.08421)] [[文章](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F10\u002Fannouncing-ava-finely-labeled-video.html)] [[数据集](https:\u002F\u002Fresearch.google.com\u002Fava\u002F)]\n- 不依赖人类知识掌握围棋游戏 [[Nature](https:\u002F\u002Fwww.nature.com\u002Farticles\u002Fnature24270.epdf?author_access_token=VJXbVjaSHxFoctQQ4p2k4tRgN0jAjWel9jnR3ZoTv0PVW4gB86EEpGqTRDtpIz-2rmo8-KG06gqVobU5NSCFeHILHcVFUeMsbvwS-lxjqQGg98faovwjxeTUgZAUMnRQ)] [[文章](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Falphago-zero-learning-scratch\u002F)]\n- 基于动力学随机化的机器人控制从仿真到现实的迁移 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.06537)] [[文章](https:\u002F\u002Fblog.openai.com\u002Fgeneralizing-from-simulation\u002F)]\n- 基于图像的机器人学习中的非对称演员评论家算法 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.06542)] [[文章](https:\u002F\u002Fblog.openai.com\u002Fgeneralizing-from-simulation\u002F)]\n- 卷积神经网络中类别不平衡问题的系统性研究 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.05381)]\n- 深度学习中的泛化能力 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.05468)]\n- Swish：一种自门控激活函数 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.05941)]\n- 多智能体通信中的涌现式翻译 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.06922)]\n- SLING：一种用于框架语义解析的框架 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.07032)] [[文章](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F11\u002Fsling-natural-language-frame-semantic.html)] [[代码](https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fsling)]\n- 摔跤领域的元学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.03641)] [[文章](https:\u002F\u002Fblog.openai.com\u002Fmeta-learning-for-wrestling\u002F)] [[代码](https:\u002F\u002Fgithub.com\u002Fopenai\u002Frobosumo)]\n- 混合精度训练 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.03740)] [[文章](http:\u002F\u002Fresearch.baidu.com\u002Fmixed-precision-training\u002F)] [[文章2](https:\u002F\u002Fdevblogs.nvidia.com\u002Fparallelforall\u002Fmixed-precision-training-deep-neural-networks\u002F)] [[文档\u002F代码](http:\u002F\u002Fdocs.nvidia.com\u002Fdeeplearning\u002Fsdk\u002Fmixed-precision-training\u002Findex.html)]\n- 生成对抗网络：概述 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.07035)]\n- 多智能体竞争中的涌现复杂性 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.03748)] [[文章](https:\u002F\u002Fblog.openai.com\u002Fcompetitive-self-play\u002F)] [[代码](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fmultiagent-competition)]\n- 深度格网网络与部分单调函数 [[Google研究](https:\u002F\u002Fresearch.google.com\u002Fpubs\u002Fpub46327.html)] [[文章](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F10\u002Ftensorflow-lattice-flexibility.html)] [[代码](https:\u002F\u002Fgithub.com\u002Ftensorflow\u002Flattice)]\n- 孟买理工学院英印平行语料库 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.02855)] [[文章](http:\u002F\u002Fwww.cfilt.iitb.ac.in\u002Fiitb_parallel\u002F)]\n- Rainbow：结合深度强化学习中的多项改进 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.02298)]\n- 基于动态可扩展网络的终身学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.01547)]\n- 变分推断与深度学习：一种新的综合（论文） [[dropbox](https:\u002F\u002Fwww.dropbox.com\u002Fs\u002Fv6ua3d9yt44vgb3\u002Fcover_and_thesis.pdf)]\n- 神经任务编程：学习在层次化任务间泛化 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.01813)]\n- 图像之间的神经色彩迁移 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1710.00756)]\n- 海马体作为预测地图 [[biorXiv](https:\u002F\u002Fwww.biorxiv.org\u002Fcontent\u002Fbiorxiv\u002Fearly\u002F2017\u002F07\u002F25\u002F097170.full.pdf)] [[文章](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fhippocampus-predictive-map\u002F)]\n- 面向电子健康记录的可扩展且准确的深度学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1801.07860)]\n\n#### 2017年9月\n\n- 生成模型中的变分记忆寻址 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.07116)]\n- 利用示范克服强化学习中的探索问题 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.10089)]\n- 一种结合数字信号处理与深度学习的实时全频段语音增强方法 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.08243)] [[文章](https:\u002F\u002Fpeople.xiph.org\u002F~jm\u002Fdemo\u002Frnnoise\u002F)] [[代码](https:\u002F\u002Fgithub.com\u002Fxiph\u002Frnnoise\u002F)]\n- ChestX-ray8：医院规模的胸部X光数据库及常见胸腔疾病的弱监督分类与定位基准测试 [[CVF](http:\u002F\u002Fopenaccess.thecvf.com\u002Fcontent_cvpr_2017\u002Fpapers\u002FWang_ChestX-ray8_Hospital-Scale_Chest_CVPR_2017_paper.pdf)] [[文章](https:\u002F\u002Fwww.nih.gov\u002Fnews-events\u002Fnews-releases\u002Fnih-clinical-center-provides-one-largest-publicly-available-chest-x-ray-datasets-scientific-community)] [[数据集](https:\u002F\u002Fnihcc.app.box.com\u002Fv\u002FChestXray-NIHCC)]\n- NIMA：神经图像评估 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.05424)] [[文章](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F12\u002Fintroducing-nima-neural-image-assessment.html)]\n- 通过编辑原型生成句子 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.08878)] [[代码](https:\u002F\u002Fgithub.com\u002Fkelvinguu\u002Fneural-editor)]\n- 意识先验 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.08568)]\n- StarSpace：将所有内容嵌入！[[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.03856)] [[代码](https:\u002F\u002Fgithub.com\u002Ffacebookresearch\u002FStarspace)]\n- 基于强化学习的神经网络优化器搜索 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.07417)]\n- 神经序列模型的动态评估 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.07432)]\n- 神经机器翻译 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.07809)]\n- Matterport3D：基于室内环境的RGB-D数据学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.06158)] [[文章](https:\u002F\u002Fniessner.github.io\u002FMatterport\u002F)] [[文章2](https:\u002F\u002Fhackernoon.com\u002Fannouncing-the-matterport3d-research-dataset-815cae932939)] [[代码](https:\u002F\u002Fgithub.com\u002Fniessner\u002FMatterport)]\n- 重要的深度强化学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.06560)] [[代码](https:\u002F\u002Fgithub.com\u002FBreakend\u002FDeepReinforcementLearningThatMatters)]\n- 不确定性贝尔曼方程与探索 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.05380)]\n- WESPE：用于数码相机的弱监督照片增强器 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.01118)] [[文章](http:\u002F\u002Fpeople.ee.ethz.ch\u002F~ihnatova\u002Fwespe.html)]\n- 全局归一化阅读器 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.02828)] [[文章](http:\u002F\u002Fresearch.baidu.com\u002Fgnr\u002F)] [[代码](https:\u002F\u002Fgithub.com\u002Fbaidu-research\u002FGloballyNormalizedReader)]\n- 工程师入门级机器学习简明介绍 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.02840)]\n- 具有对手学习意识的学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.04326)] [[文章](https:\u002F\u002Fblog.openai.com\u002Flearning-to-model-other-minds\u002F)]\n- 一款深度强化学习聊天机器人 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.02349)]\n- 激励挤压网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1709.01507)]\n- 深度学习的高效方法与硬件（论文） [[斯坦福数字资源库](https:\u002F\u002Fpurl.stanford.edu\u002Fqf934gh3708)]\n\n#### 2017年8月\n\n- NIPS 2016 审稿流程的设计与分析 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.09794)]\n- 使用卷积神经网络对强引力透镜进行快速自动化分析 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.08842)] [[文章](http:\u002F\u002Fwww.symmetrymagazine.org\u002Farticle\u002Fneural-networks-meet-space)]\n- TensorFlow Agents：在 TensorFlow 中实现高效的批量强化学习 [[白皮书](https:\u002F\u002Fdrive.google.com\u002Ffile\u002Fd\u002F0B20Yn-GSaVHGMVlPanRTRlNIRlk\u002Fview)] [[代码](https:\u002F\u002Fgithub.com\u002Ftensorflow\u002Fagents)]\n- 在在线评论系统中自动化的众包刷评攻击与防御 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.08151)]\n- 基于模型的深度强化学习结合无模型微调的神经网络动力学 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.02596)] [[文章](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2017\u002F11\u002F30\u002Fmodel-based-rl\u002F)] [[代码](https:\u002F\u002Fgithub.com\u002Fnagaban2\u002Fnn_dynamics)]\n- 深度学习在视频游戏中的应用 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.07902)]\n- 用于广告点击预测的深度交叉网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.05123)]\n- Fashion-MNIST：一个用于机器学习算法基准测试的新图像数据集 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.07747)] [[代码](https:\u002F\u002Fgithub.com\u002Fzalandoresearch\u002Ffashion-mnist)]\n- 多任务自监督视觉学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.07860)]\n- 学习多视角立体视觉系统 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.05375)] [[文章](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2017\u002F09\u002F05\u002Funified-3d\u002F)] [[代码]()]（即将发布）\n- 双生网络：利用未来作为正则化手段 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.06742)]\n- 深度强化学习简述 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.05866)]\n- 基于克罗内克分解近似的可扩展信任区域方法，用于深度强化学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.05144)] [[代码](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fbaselines)]\n- 可见水印的有效性研究 [[CVPR](http:\u002F\u002Fopenaccess.thecvf.com\u002Fcontent_cvpr_2017\u002Fpapers\u002FDekel_On_the_Effectiveness_CVPR_2017_paper.pdf)] [[文章](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F08\u002Fmaking-visible-watermarks-more-effective.html)]\n- 基于 Q 学习的实用网络模块设计 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.05552)]\n- 关于确保智能机器行为规范的研究 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.05448)]\n- 连续控制领域深度强化学习基准任务的可重复性 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.04133)] [[代码](https:\u002F\u002Fgithub.com\u002FBreakend\u002FReproducibilityInContinuousPolicyGradientMethods)]\n- 使用深度强化学习训练深度自编码器进行协同过滤 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.01715)] [[代码](https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FDeepRecommender)]\n- 利用深度强化学习学习如何在地面上完成停靠式着陆 [[nature](https:\u002F\u002Flink.springer.com\u002Fepdf\u002F10.1007\u002Fs10846-017-0696-1?author_access_token=BEvJgzY3QauUddBuQAus2ve4RwlQNchNByi7wbcMAY5xhRRqI6HVNnXt8Pgp850SnuV5ue6mUo3Jc7FIP5FgLmqk34Wob3oqyuGtkg7E_1T0dg02IYhfY-3dvb8R9zEmaGzTogYCIXm4O4vZ_tSGnA%3D%3D)]\n- 重新审视现成时序建模方法在大规模视频分类中的有效性 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.03805)] [[文章](http:\u002F\u002Fresearch.baidu.com\u002Fspatial-temporal-modeling-framework-large-scale-video-understanding\u002F)]\n- 具有自动课程学习的内在动机目标探索过程 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.02190)]\n- 神经期望最大化 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.03498)] [[代码](https:\u002F\u002Fgithub.com\u002Fsjoerdvansteenkiste\u002F)]\n- Google Vizier：黑箱优化服务 [[Google 研究](https:\u002F\u002Fresearch.google.com\u002Fpubs\u002Fpub46180.html)]\n- STARDATA：星际争霸 AI 研究数据集 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.02139)] [[代码](https:\u002F\u002Fgithub.com\u002FTorchCraft\u002FStarData)]\n- 利用数百万个表情符号实例学习跨领域的情感、情绪和讽刺检测表示 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.00524)] [[代码](https:\u002F\u002Fgithub.com\u002Fbfelbo\u002Fdeepmoji)] [[文章](https:\u002F\u002Fwww.media.mit.edu\u002Fposts\u002Fwhat-can-we-learn-from-emojis\u002F)]\n- 使用小型前馈网络进行自然语言处理 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.00214)]\n\n#### 2017年7月\n\n- 基于级联精炼网络的摄影图像合成 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.09405)] [[代码](https:\u002F\u002Fgithub.com\u002FCQFIO\u002FPhotographicImageSynthesis)]\n- 星际争霸II：强化学习的新挑战 [[DeepMind文档](https:\u002F\u002Fdeepmind.com\u002Fdocuments\u002F110\u002Fsc2le.pdf)] [[代码](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Fpysc2)] [[文章](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fdeepmind-and-blizzard-open-starcraft-ii-ai-research-environment\u002F)]\n- 利用示范在稀疏奖励的机器人问题上进行深度强化学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.08817)]\n- 基于深度能量函数策略的强化学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1702.08165)] [[文章](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2017\u002F10\u002F06\u002Fsoft-q-learning\u002F)] [[代码](https:\u002F\u002Fgithub.com\u002Fhaarnoja\u002Fsoftqlearning)]\n- DARLA：提升强化学习中的零样本迁移能力 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.08475)]\n- 合成鲁棒的对抗样本 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.07397)] [[文章](http:\u002F\u002Fwww.labsix.org\u002Fphysical-objects-that-fool-neural-nets\u002F)] [[代码]()]（即将发布）\n- 通过语音循环为野生环境中的说话者进行语音合成 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.06588)] [[代码](https:\u002F\u002Fgithub.com\u002Ffacebookresearch\u002Floop)] [[文章](https:\u002F\u002Fytaigman.github.io\u002Floop\u002F)]\n- Eyemotion：利用眼动追踪摄像头在VR中分类面部表情 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.07204)] [[文章](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F07\u002Fexpressions-in-virtual-reality.html)]\n- 强化学习的分布视角 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.06887)] [[文章](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fgoing-beyond-average-reinforcement-learning\u002F)] [[视频](https:\u002F\u002Fvimeo.com\u002F235922311)]\n- 神经语言模型评估的现状 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.05589)]\n- 优化生成网络的潜在空间 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.05776)]\n- 受神经科学启发的人工智能 [[Neuron](http:\u002F\u002Fwww.cell.com\u002Fneuron\u002Ffulltext\u002FS0896-6273(17)30509-3?_returnURL=http%3A%2F%2Flinkinghub.elsevier.com%2Fretrieve%2Fpii%2FS0896627317305093%3Fshowall%3Dtrue)] [[文章](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fai-and-neuroscience-virtuous-circle\u002F)]\n- 学习可迁移的架构以实现可扩展的图像识别 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.07012)]\n- 强化学习中的逆向课程生成 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.05300)]\n- 基于想象力增强的智能体用于深度强化学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.06203)] [[文章](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fagents-imagine-and-plan\u002F)]\n- 从零开始学习基于模型的规划 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.06170)] [[文章](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fagents-imagine-and-plan\u002F)]\n- 近端策略优化算法 [[AWSS3](https:\u002F\u002Fopenai-public.s3-us-west-2.amazonaws.com\u002Fblog\u002F2017-07\u002Fppo\u002Fppo-arxiv.pdf)] [[代码](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fbaselines)]\n- 自动识别欺骗性情绪面部表情 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.04061)]\n- Distral：鲁棒的多任务强化学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.04175)]\n- Creatism：一位能够创作专业作品的深度学习摄影师 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.03491)] [[文章](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F07\u002Fusing-deep-learning-to-create.html)]\n- SCAN：学习抽象的分层组合视觉概念 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.03389)] [[文章](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fimagine-creating-new-visual-concepts-recombining-familiar-ones\u002F)]\n- 重新审视大数据在深度学习时代的不合理有效性 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.02968)] [[文章](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F07\u002Frevisiting-unreasonable-effectiveness.html)]\n- 故意的非故意智能体：同时学习解决多个连续控制任务 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.03300)]\n- 用于实时图像增强的深度双边学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.02880)] [[代码](https:\u002F\u002Fgithub.com\u002Fmgharbi\u002Fhdrnet)] [[文章](https:\u002F\u002Fgroups.csail.mit.edu\u002Fgraphics\u002Fhdrnet\u002F)]\n- 在丰富环境中涌现的运动行为 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.02286)] [[文章](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fproducing-flexible-behaviours-simulated-environments\u002F)]\n- 通过对抗模仿从动作捕捉数据中学习人类行为 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.02201)] [[文章](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fproducing-flexible-behaviours-simulated-environments\u002F)]\n- 多样化行为的鲁棒模仿 [[arXiv](https:\u002F\u002Fdeepmind.com\u002Fdocuments\u002F95\u002Fdiverse_arxiv.pdf)] [[文章](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fproducing-flexible-behaviours-simulated-environments\u002F)]\n- [事后经验回放](notes\u002Fhindsight-ep.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.01495)]\n- 使用卷积神经网络实现心脏病专家级别的心律失常检测 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.01836)] [[文章](https:\u002F\u002Fstanfordmlgroup.github.io\u002Fprojects\u002Fecg\u002F)]\n- 语义抓取的端到端学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.01932)]\n- ELF：一个广泛、轻量且灵活的即时战略游戏研究平台 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1707.01067)] [[代码](https:\u002F\u002Fgithub.com\u002Ffacebookresearch\u002FELF)] [[文章](https:\u002F\u002Fcode.facebook.com\u002Fposts\u002F132985767285406\u002Fintroducing-elf-an-extensive-lightweight-and-flexible-platform-for-game-research\u002F)]\n\n#### 2017年6月\n\n- [用于探索的噪声网络](notes\u002Fnoisy-networks-4-exploration.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.10295)]\n- GAN 真的能学习到数据分布吗？一项实证研究 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.08224)]\n- 用于连续学习的梯度剧集记忆 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.08840)]\n- 多智能体对话中自然语言并不会“自然”出现 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.08502)] [[代码](https:\u002F\u002Fgithub.com\u002Fbatra-mlp-lab\u002Flang-emerge)]\n- 用于点击率预估的深度兴趣网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.06978)]\n- 面向深度神经网络的认知心理学：以形状偏见为例 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.08606)] [[文章](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fcognitive-psychology\u002F)]\n- 运动控制中的结构学习：一种深度强化学习模型 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.06827)]\n- 可编程智能体 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.06383)]\n- 在模拟 3D 世界中的具身语言学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.06551)]\n- 模式网络：基于生成因果模型的直觉物理零样本迁移 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.04317)]\n- SVCCA：用于深度学习动态与可解释性的奇异向量典型相关分析 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.05806)] [[文章](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F11\u002Finterpreting-deep-neural-networks-with.html)] [[代码](https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fsvcca)]\n- 一个模型即可学会所有任务 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.05137)] [[代码](https:\u002F\u002Fgithub.com\u002Ftensorflow\u002Ftensor2tensor)] [[文章](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F06\u002Fmultimodel-multi-task-machine-learning.html)]\n- 强化学习中的混合奖励架构 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.04208)]\n- 期望策略梯度 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.05374)]\n- 用于自编码生成对抗网络的变分方法 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.04987)]\n- 谈判还是不谈？面向谈判对话的端到端学习 [[S3AWS](https:\u002F\u002Fs3.amazonaws.com\u002Fend-to-end-negotiator\u002Fend-to-end-negotiator.pdf)] [[代码](https:\u002F\u002Fgithub.com\u002Ffacebookresearch\u002Fend-to-end-negotiator)] [[文章](https:\u002F\u002Fcode.facebook.com\u002Fposts\u002F1686672014972296\u002Fdeal-or-no-deal-training-ai-bots-to-negotiate\u002F)]\n- 注意力就是你所需要的 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.03762)] [[代码](https:\u002F\u002Fgithub.com\u002Ftensorflow\u002Ftensor2tensor)] [[文章](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F08\u002Ftransformer-novel-neural-network.html)]\n- 神经网络的索博列夫训练 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.04859)]\n- YellowFin 与动量调优的艺术 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.03471)] [[代码](https:\u002F\u002Fgithub.com\u002FJianGoForIt\u002FYellowFin)] [[文章](http:\u002F\u002Fdawn.cs.stanford.edu\u002F2017\u002F07\u002F05\u002Fyellowfin\u002F)]\n- 前瞻性思维：逐层构建和训练神经网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.02480)]\n- 用于神经机器翻译的深度可分离卷积 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.03059)] [[代码](https:\u002F\u002Fgithub.com\u002Ftensorflow\u002Ftensor2tensor)]\n- 用于探索的参数空间噪声 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.01905)] [[代码](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fbaselines)] [[文章](https:\u002F\u002Fblog.openai.com\u002Fbetter-exploration-with-parameter-noise\u002F)]\n- 基于人类偏好的深度强化学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.03741)] [[文章](https:\u002F\u002Fblog.openai.com\u002Fdeep-reinforcement-learning-from-human-preferences\u002F)]\n- 用于混合合作-竞争环境的多智能体演员-评论家算法 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.02275)] [[代码](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fmultiagent-particle-envs)]\n- 自归一化神经网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.02515)] [[代码](https:\u002F\u002Fgithub.com\u002Fbioinf-jku\u002FSNNs)]\n- 高精度、大批次 SGD：1 小时内训练 ImageNet [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.02677)]\n- 一种用于关系推理的简单神经网络模块 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.01427)] [[文章](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fneural-approach-relational-reasoning\u002F)]\n- 视觉交互网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1706.01433)] [[文章](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Fneural-approach-relational-reasoning\u002F)]\n\n#### 2017年5月\n\n- 从自然语言推理数据中监督学习通用句子表示 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.02364)]  [[代码](https:\u002F\u002Fgithub.com\u002Ffacebookresearch\u002FInferSent)]\n- pix2code：从图形用户界面截图生成代码 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.07962)] [[文章](https:\u002F\u002Fuizard.io\u002Fresearch#pix2code)] [[代码](https:\u002F\u002Fgithub.com\u002Ftonybeltramelli\u002Fpix2code)]\n- 克拉默距离作为解决偏置Wasserstein梯度问题的方案 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.10743)]\n- 奖励通道受损情况下的强化学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.08417)]\n- 空洞残差网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.09914)] [[代码](https:\u002F\u002Fgithub.com\u002Ffyu\u002Fdrn)]\n- 贝叶斯GAN [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.09558)] [[代码](https:\u002F\u002Fgithub.com\u002Fandrewgordonwilson\u002Fbayesgan\u002F)]\n- 梯度下降可能需要指数时间才能逃离鞍点 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.10412)] [[文章](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2017\u002F08\u002F31\u002Fsaddle-efficiency\u002F)]\n- 使用深度学习和树搜索进行快速与慢速思考 [[arXiv]()]\n- ParlAI：对话研究软件平台 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.06476)] [[代码](https:\u002F\u002Fgithub.com\u002Ffacebookresearch\u002FParlAI)] [[文章](https:\u002F\u002Fcode.facebook.com\u002Fposts\u002F266433647155520\u002Fparlai-a-new-software-platform-for-dialog-research\u002F)]\n- 语义分解生成对抗网络的潜在空间 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.07904)] [[文章](https:\u002F\u002Faws.amazon.com\u002Fblogs\u002Fai\u002Fcombining-deep-learning-networks-gan-and-siamese-to-generate-high-quality-life-like-images\u002F)]\n- 看、听并学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.08168)]\n- 行动识别何去何从？一种新模型及Kinetics数据集 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.07750)] [[代码](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Fkinetics-i3d)]\n- 卷积序列到序列学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.03122)] [[代码](https:\u002F\u002Fgithub.com\u002Ffacebookresearch\u002Ffairseq)] [[代码2](https:\u002F\u002Fgithub.com\u002Ffacebookresearch\u002Ffairseq-py)] [[文章](https:\u002F\u002Fcode.facebook.com\u002Fposts\u002F1978007565818999\u002Fa-novel-approach-to-neural-machine-translation\u002F)]\n- Kinetics人类动作视频数据集 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.06950)] [[文章](https:\u002F\u002Fdeepmind.com\u002Fresearch\u002Fopen-source\u002Fopen-source-datasets\u002Fkinetics\u002F)]\n- 不完全信息博弈中的安全与嵌套子博弈求解 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.02955)]\n- 针对深度强化学习的连续动作离散序列预测 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.05035)]\n- 用于自适应想象优化的元控制 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.02670)]\n- 针对深度强化学习的高效并行方法 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.04862)]\n- 实时自适应图像压缩 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1705.05823)]\n\n#### 2017年4月\n\n- 通用视频游戏AI：从屏幕截图中学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.06945)]\n- 学习略读文本 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.06877)]\n- 抓住要点：基于指针-生成器网络的摘要生成 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.04368)] [[代码](https:\u002F\u002Fgithub.com\u002Fabisee\u002Fpointer-generator)] [[文章](http:\u002F\u002Fwww.abigailsee.com\u002F2017\u002F04\u002F16\u002Ftaming-rnns-for-better-summarization.html)]\n- 对抗性神经机器翻译 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.06933)]\n- [从示范中进行深度Q学习](notes\u002Fdqn-demonstrations.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.03732)]\n- 针对现实世界强化学习的示范学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.03732)]\n- 使用深度卷积网络在移动设备上拍摄单反相机质量的照片 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.02470)] [[文章](http:\u002F\u002Fpeople.ee.ethz.ch\u002F~ihnatova\u002F)] [[代码](https:\u002F\u002Fgithub.com\u002Faiff22\u002FDPED)]\n- 草图绘制的神经网络表示 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.03477)] [[代码](https:\u002F\u002Fgithub.com\u002Ftensorflow\u002Fmagenta\u002Ftree\u002Fmaster\u002Fmagenta\u002Fmodels\u002Fsketch_rnn)] [[文章](https:\u002F\u002Fresearch.googleblog.com\u002F2017\u002F04\u002Fteaching-machines-to-draw.html)]\n- 针对神经网络的自动化课程学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.03003)]\n- 用于3D物体重建的分层表面预测 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.00710)] [[文章](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2017\u002F08\u002F23\u002Fhigh-quality-3d-obj-reconstruction\u002F)]\n- 用于量子化学的神经消息传递 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.01212)]\n- 学习生成评论并发现情感 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.01444)] [[代码](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fgenerating-reviews-discovering-sentiment)]\n- 将深度学习应用于新兴领域的最佳实践 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.01568)]\n\n#### 2017年3月\n\n- 改进的 Wasserstein GAN 训练 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1704.00028)]\n- 进化策略：强化学习的可扩展替代方案 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.03864)]\n- 可控文本生成 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.00955)]\n- 神经情景控制 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.01988)]\n- [一种结构化的自注意力句子嵌入](notes\u002Fself_attention_embedding.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.03130)]\n- 多步强化学习：一种统一的算法 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.01327)]\n- 使用卷积神经网络进行深度学习，用于脑图谱绘制及从人类脑电图中解码运动相关信息 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.05051)]\n- FaSTrack：用于快速且保证安全的运动规划的模块化框架 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.07373)] [[文章](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2017\u002F12\u002F05\u002Ffastrack\u002F)] [[文章2](http:\u002F\u002Fsylviaherbert.com\u002Ffastrack\u002F)]\n- 大规模探索神经机器翻译架构 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.03906)] [[代码](https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fseq2seq)]\n- 通过直接体素 CNN 回归从单张图像重建大姿态三维人脸 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.07834)] [[文章](http:\u002F\u002Faaronsplace.co.uk\u002Fpapers\u002Fjackson2017recon\u002F)] [[代码](https:\u002F\u002Fgithub.com\u002FAaronJackson\u002Fvrn)]\n- 强化学习的极小极大后悔界 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.05449)]\n- 锐利的极小值可以泛化到深度网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.04933)]\n- 并行多尺度自回归密度估计 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.03664)]\n- 神经机器翻译与序列到序列模型：教程 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.01619)]\n- 图像分类器的大规模进化 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.01041)]\n- 用于层次化强化学习的封建网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.01161)]\n- 深度神经网络的进化 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.00548)]\n- 如何高效地逃离鞍点 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.00887)] [[文章](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2017\u002F08\u002F31\u002Fsaddle-efficiency\u002F)]\n- 通过信息打开深度神经网络的黑箱 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.00810)] [[视频](https:\u002F\u002Fyoutu.be\u002FbLqJHjXihK8)]\n- 理解合成梯度与解耦神经接口 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.00522)]\n- 学习优化神经网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1703.00441)] [[文章](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2017\u002F09\u002F12\u002Flearning-to-optimize-with-rl\u002F)]\n\n\n#### 2017年2月\n\n- 破碎梯度问题：如果 ResNet 是答案，那么问题是什么？[[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1702.08591)]\n- 神经地图：深度强化学习中的结构化记忆 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1702.08360)]\n- 拉近基于价值和基于策略的强化学习之间的差距 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1702.08892)]\n- Deep Voice：实时神经文本转语音 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1702.07825)]\n- 使用深度强化学习击败《任天堂明星大乱斗》世界冠军 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1702.06230)]\n- 游戏模仿：用于快速视频游戏 AI 的深度监督卷积网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1702.05663)]\n- 学习解析与翻译能提升神经机器翻译性能 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1702.03525)]\n- 除顶级之外的一切：词表示的简单而有效的后处理 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1702.01417)]\n- 使用动态计算图进行深度学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1702.02181)]\n- 跳跃连接作为有效的对称破缺 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1701.09175)]\n- 基于生成式领域自适应网络的半监督问答模型 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1702.02206)]\n\n#### 2017年1月\n\n- Wasserstein GAN [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1701.07875)]\n- 需求深度强化学习：概述 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1701.07274)]\n- DyNet：动态神经网络工具包 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1701.03980)]\n- DeepStack：无限制扑克中的专家级人工智能 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1701.01724)]\n- NIPS 2016 教程：生成对抗网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1701.00160)]\n\n#### 2016年12月\n\n- [一种没有混沌的循环神经网络](notes\u002Frnn_no_chaos.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1612.06212)]\n- 使用门控卷积网络进行语言建模 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1612.08083)]\n- EnhanceNet：通过自动纹理合成实现单幅图像超分辨率 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1612.07919)] [[文章](http:\u002F\u002Fwebdav.tuebingen.mpg.de\u002Fpixel\u002Fenhancenet\u002F)]\n- 通过对抗训练从模拟和无监督图像中学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1612.07828)]\n- 字符级神经机器翻译有多语法？用对比翻译对评估 MT 质量 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1612.04629)]\n- 使用连续缓存改进神经语言模型 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1612.04426)]\n- DeepMind 实验室 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1612.03801)] [[代码](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Flab)]\n- 在强弱人类监督下，无需模拟器即可学习机器人任务的深度学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1612.01086)]\n- 知道何时该看：通过视觉哨兵实现图像字幕的自适应注意力 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1612.01887)]\n- 克服神经网络中的灾难性遗忘 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1612.00796)]\n\n#### 2016年11月（ICLR 版）\n\n- 基于条件对抗网络的图像到图像翻译 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.07004)]\n- [极其庞大的神经网络：稀疏门控专家混合层](notes\u002Fmixture-experts.md) [[OpenReview](https:\u002F\u002Fopenreview.net\u002Fforum?id=B1ckMDqlg)]\n- 学习如何进行强化学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.05763)]\n- 奥德赛的出路：分析并结合LSTM的最新见解 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.05104)]\n- [半监督文本分类的对抗训练方法](notes\u002Fadversarial-text-classification.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1605.07725)]\n- 不等支持下的重要性采样 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.03451)]\n- 准循环神经网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.01576)]\n- 循环神经网络中的容量与可学习性 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=BydARw9ex)]\n- 展开式生成对抗网络 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=BydrOIcle)]\n- 深度信息传播 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=H1W1UN9gg)]\n- 结构化注意力网络 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=HkE0Nvqlg)]\n- 增量序列学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.03068)]\n- 探讨可迁移的对抗样本与黑盒攻击 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.02770)] [[代码](https:\u002F\u002Fgithub.com\u002FReDeiPirati\u002Ftransferability-advdnn-pub)]\n- b-GAN：生成对抗网络的统一框架 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=S1JG13oee)]\n- 一种多任务联合模型：为多个自然语言处理任务生长神经网络 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=SJZAb5cel)]\n- 使用Gumbel-Softmax进行类别重参数化 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.01144)]\n- 在野外唇读句子 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.05358)]\n\n强化学习：\n\n- 学习如何进行强化学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.05763)]\n- 生成对抗网络、逆向强化学习和基于能量的模型之间的联系 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.03852)]\n- 预测机：端到端学习与规划 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=BkJsCIcgl)]\n- [第三人称模仿学习](notes\u002Fthird-person-imitation-learning.md) [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=B16dGcqlx)]\n- 通过半监督强化学习泛化技能 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=ryHlUtqge)]\n- 具有经验回放的高效演员-评论家算法 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=HyM25Mqel)]\n- [带有无监督辅助任务的强化学习](notes\u002Frl-auxiliary-tasks.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.05397)]\n- 使用强化学习进行神经架构搜索 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=r1Ue8Hcxg)]\n- 向信息寻求型智能体迈进 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=SyW2QSige)]\n- 多智能体合作与（自然）语言的涌现 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=Hk8N3Sclg)]\n- 通过探索被低估的奖励来改进策略梯度 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=ryT4pvqll)]\n- 用于层次化强化学习的随机神经网络 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=B1oK8aoxe)]\n- 使用强化学习调优循环神经网络 [[OpenReview](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.02796)]\n- RL^2：通过慢速强化学习实现快速强化学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.02779)]\n- 学习不变特征空间以通过强化学习迁移技能 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=Hyq4yhile)]\n- 通过深度强化学习学习执行物理实验 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=r1nTpv9eg)]\n- 在GPU上使用异步优势演员-评论家算法进行强化学习 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=r1VGvBcxl)]\n- 学习用强化学习将单词组合成句子 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=Skvgqgqxe)]\n- 用于加速收敛速度的深度强化学习 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=Syg_lYixe)]\n- [#探索：关于深度强化学习中基于计数的探索研究](notes\u002Fcount-based-exploration.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.04717)]\n- 学习用强化学习将单词组合成句子 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=Skvgqgqxe)]\n- 学习在复杂环境中导航 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.03673)]\n- 用于模仿学习的无监督感知奖励 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=Bkul3t9ee)]\n- Q-Prop：具有离策略评论家的高效策略梯度 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=SJ3rcZcxl)]\n\n\n机器翻译与对话\n\n- [谷歌的多语言神经机器翻译系统：实现零样本翻译](notes\u002Fgnmt-multilingual.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.04558)]\n- [基于重建的神经机器翻译](notes\u002Fnmt-with-reconstruction.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.01874v1)]\n- 机器翻译的迭代优化 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=r1y1aawlg)]\n- 用于神经机器翻译的卷积编码器模型 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.02344)]\n- 通过连续缓存改进神经语言模型 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=B184E5qee)]\n- 神经机器翻译中的词汇选择策略 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=Bk8N0RLxx)]\n- 向自动图灵测试迈进：学习评估对话回复 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=HJ5PIaseg)]\n- 人机协作的对话学习 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=HJgXCV9xx)]\n- 批量策略梯度方法用于改进神经对话模型 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=rJfMusFll)]\n- 通过对话交互学习 [[OpenReview](http:\u002F\u002Fopenreview.net\u002Fforum?id=rkE8pVcle)]\n- [机器翻译的对偶学习](notes\u002Fdual-learning-mt.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.00179)]\n- 用于序列到序列学习的无监督预训练 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.02683)]\n\n\n\n#### 2016年10月\n\n- 使用具有动态外部记忆的神经网络的混合计算 [[nature](https:\u002F\u002Fwww.nature.com\u002Farticles\u002Fnature20101.epdf?author_access_token=ImTXBI8aWbYxYQ51Plys8NRgN0jAjWel9jnR3ZoTv0MggmpDmwljGswxVdeocYSurJ3hxupzWuRNeGvvXnoO8o4jTJcnAyhGuZzXJ1GEaD-Z7E6X_a9R-xqJ9TfJWBqz)] [[code](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Fdnc)]\n- 量子机器学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.09347)]\n- 理解深度学习需要重新思考泛化能力 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1611.03530)]\n- 普适性对抗扰动 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.08401)] [[code](https:\u002F\u002Fgithub.com\u002FLTS4\u002Funiversal)]\n- [线性时间下的神经机器翻译](notes\u002Fnmt-linear-time.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.10099)] [[code](https:\u002F\u002Fgithub.com\u002Ftensorflow\u002Ftensor2tensor)]\n- [教授强制：训练循环网络的新算法](notes\u002Fprofessor-forcing.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.09038)]\n- 学习用对抗性神经密码学保护通信 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.06918v1)]\n- 主动记忆能否取代注意力机制？[[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.08613)]\n- [利用快速权重关注近期历史](notes\u002Ffast-weight-to-attend.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.06258)]\n- [无需显式分词的全字符级神经机器翻译](notes\u002Fconv-char-level-nmt.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.03017)]\n- [多样化束搜索：从神经序列模型中解码多样化的解决方案](notes\u002Fdiverse-beam-search.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.02424)]\n- 视频像素网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.00527)]\n- 将生成对抗网络与演员-评论家方法连接起来 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.01945)]\n- [使用神经机器翻译实时学习翻译](notes\u002Flearning-to-translate-real-time.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.00388)]\n- Xception：基于深度可分离卷积的深度学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.02357)]\n- 基于分布式异步引导策略搜索的群体机器人强化学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1610.00673)]\n- [指针哨兵混合模型](notes\u002Fpointer-sentinel-mixture.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1609.07843)]\n\n#### 2016年9月\n\n- 朝向深度符号强化学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1609.05518)]\n- 超网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1609.09106)]\n- 谷歌的神经机器翻译系统：弥合人类与机器翻译之间的鸿沟 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1609.08144)]\n- 安全高效的离策略强化学习 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.02647)]\n- 使用深度强化学习玩FPS游戏 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1609.05521)]\n- [SeqGAN：带有策略梯度的序列生成对抗网络](notes\u002Fseq-gan.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1609.05473)]\n- 针对深度确定性策略的周期性探索：应用于星际争霸微观操作任务 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1609.02993)]\n- 基于能量的生成对抗网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1609.03126)]\n- 通过预测API窃取机器学习模型 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1609.02943)]\n- 基于图卷积网络的半监督分类 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1609.02907)]\n- WaveNet：一种用于原始音频的生成模型 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1609.03499)]\n- [层次多尺度循环神经网络](notes\u002Fhm-rnn.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1609.01704)]\n- 用于信息获取的对话代理端到端强化学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1609.00777)]\n- 用于YouTube推荐的深度神经网络 [[paper](https:\u002F\u002Fresearch.google.com\u002Fpubs\u002Fpub45530.html)]\n\n#### 2016年8月\n\n- 自动从语言语料库中提取的语义包含类似人类的偏见 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1608.07187)]\n- 为什么深度且廉价的学习效果如此好？[[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1608.08225)]\n- 使用匹配LSTM和答案指针的机器阅读理解 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1608.07905)]\n- 堆叠近似回归机：一种简单的深度学习方法 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1608.04062)]\n- 使用合成梯度解耦的神经接口 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1608.05343)]\n- WikiReading：一项基于维基百科的新型大规模语言理解任务 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1608.03542)]\n- 用于神经机器翻译的时序注意力模型 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1608.02927)]\n- 残差网络的残差网络：多层级残差网络 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1608.02908)]\n- [利用连续奖励策略梯度在线学习对齐](notes\u002Fonline-alignments-pg.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1608.01281)]\n\n#### 2016年7月\n\n- [用于序列预测的演员-评论家算法](notes\u002Factor-critic-sequence.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1607.07086)]\n- 人工智能时代的认知科学：逆向工程婴儿语言学习者的路线图 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1607.08723v1)]\n- [循环神经网络的神经机器翻译](notes\u002Frecurrent-nmt.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1607.08725)]\n- MS-Celeb-1M：一个用于大规模人脸识别的数据集和基准测试 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1607.08221)]\n- [层归一化](notes\u002Flayer-norm.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1607.06450)]\n- [带有循环注意力建模的神经机器翻译](notes\u002Fnmt-rec-attention.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1607.05108)]\n- 神经语义编码器 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1607.04315)]\n- [用于阅读理解的注意力之上再加注意力的神经网络](notes\u002Fatt-over-att.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1607.04423)]\n- sk_p：MOOCs的神经程序纠正器 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1607.02902)]\n- 循环高速公路网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1607.03474)]\n- 用于高效文本分类的技巧大全 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1607.01759)]\n- 面向神经机器翻译的上下文相关词表示 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1607.00578)]\n- 具有软硬寻址方案的动态神经图灵机 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1607.00036)]\n\n#### 2016年6月\n\n- 序列到序列学习作为束搜索优化 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.02960)]\n- [序列级知识蒸馏](notes\u002Fseq-knowledge-distillation.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.07947)]\n- 用于对话系统的两阶段训练策略网络 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.03152)]\n- 向深度学习与神经科学的融合迈进 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.03813)]\n- 关于循环神经网络中的乘法整合 [[arxiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.06630)]\n- [推荐系统的宽而深学习](wide-and-deep.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.07792)]\n- 在线与离线手写汉字识别 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.05763)]\n- 变分自编码器教程 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.05908)]\n- 人工智能安全中的具体问题 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.06565)]\n- 深度强化学习发现内部模型 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.05174v1)]\n- [SQuAD：用于文本机器理解的10万+个问题](notes\u002Fsquad.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.05250)]\n- 基于PixelCNN解码器的条件图像生成 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.05328)]\n- 无模型的周期性控制 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.04460)]\n- [渐进式神经网络](notes\u002Fprogressive-nn.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.04671)]\n- 训练GAN的改进技术 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.03498)] [[代码](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fimproved-gan)]\n- 内存高效的随时间反向传播 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.03401)]\n- InfoGAN：通过信息最大化生成对抗网络实现可解释的表征学习 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.03657)]\n- 多语言神经机器翻译下的零资源翻译 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.04164)]\n- 用于直接阅读文档的键值记忆网络 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.03126)]\n- 具有快速前向连接的深度循环模型用于神经机器翻译 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.04199)]\n- 通过梯度下降学习梯度下降 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.04474)]\n- 通过交互学习语言游戏 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.02447)]\n- Zoneout：通过随机保留隐藏激活来正则化RNNs [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.01305)]\n- 智能回复：电子邮件的自动回复建议 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.04870)]\n- 用于半监督文本分类的虚拟对抗训练 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1605.07725)]\n- 用于对话生成的深度强化学习 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.01541)]\n- 用于自然语言处理的超深卷积网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.01781)]\n- 用于开放域话语连贯性的神经网络模型 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.01545)]\n- 用于细粒度实体类型分类的神经架构 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.01341)]\n- 用于一次学习的匹配网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.04080)]\n- 协作式逆强化学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.03137)] [[文章](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2017\u002F08\u002F17\u002Fcooperatively-learning-human-values\u002F)]\n- 用于文本理解的门控注意力阅读器 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.01549)]\n- [基于LSTM的端到端对话控制，通过监督和强化学习优化](notes\u002Fe2e-dialog-control-sl-rl.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.01269)]\n- 用于机器阅读的迭代交替神经注意力 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.02245)]\n- 用于神经机器翻译的记忆增强解码器 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1606.02003)]\n- 多分辨率循环神经网络：应用于对话响应生成 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.00776)]\n- 学习优化 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.01885)] [[文章](http:\u002F\u002Fbair.berkeley.edu\u002Fblog\u002F2017\u002F09\u002F12\u002Flearning-to-optimize-with-rl\u002F)]\n- [使用EpiReader进行自然语言理解](notes\u002Fepireader.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.02270)]\n- 对话上下文线索：个性化和历史在响应排序中的作用 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.00372)]\n- 对抗性学习推理 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.00704)]\n- OpenAI Gym [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.01540)] [[代码](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Flab)]\n- 用于语法错误修正的神经网络翻译模型 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1606.00189)]\n\n#### 2016年5月\n\n- 层次记忆网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1605.07427)]\n- 深度API学习 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.08535)]\n- 宽残差网络 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.07146)]\n- TensorFlow：一个用于大规模机器学习的系统 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.08695)]\n- 使用双向LSTM模型和内部注意力学习自然语言推理 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.09090)]\n- 基于深度记忆网络的方面级情感分类 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.08900)]\n- 分形网：无需残差的超深度神经网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1605.07648)]\n- 学习端到端的目标导向对话 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.07683)]\n- 基于记忆增强神经网络的一次学习 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.06065)]\n- 无不良局部极小值的深度学习 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.07110)]\n- AVEC 2016 - 抑郁、情绪和情感识别研讨会及挑战赛 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1605.01600)]\n- 数据编程：快速创建大型训练集 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.07723)]\n- 深度融合网络 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.07716)]\n- 深度投资组合理论 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.07230)]\n- 通过视频预测进行物理交互的无监督学习 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.07157)]\n- 电影描述 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1605.03705)]\n\n\n#### 2016年4月\n\n- 高阶循环神经网络 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1605.00064)]\n- 端到端手写段落识别中的联合线段分割与文本转录 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1604.08352)]\n- 分层深度强化学习：整合时间抽象与内在动机 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1604.06057)]\n- IBM 2016年英语会话式电话语音识别系统 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1604.08242)]\n- 基于对话的语言学习 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1604.06045)]\n- 使用双向长短期记忆模型和辅助损失的多语言词性标注 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1604.05529)]\n- 句子级别的语法错误识别作为序列到序列的修正 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1604.04677)]\n- 基于网络的端到端可训练任务导向型对话系统 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1604.04562)]\n- 视觉叙事 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1604.03968)]\n- 通过稳定性训练提升深度神经网络的鲁棒性 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1604.04326)]\n- [弥合残差学习、循环神经网络与视觉皮层之间的差距](notes\u002Fbridging-gap-resnet-rnn.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1604.03640)]\n- 扫描、注意与阅读：基于MDLSTM注意力的端到端手写段落识别 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1604.03286)]\n- [句子级递归主题模型：让主题自己说话](notes\u002Fslrtm.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1604.02038)]\n- [利用混合词-字符模型实现开放词汇量的神经机器翻译](notes\u002Fopen-vocab-nmt-hybrid-word-character.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1604.00788)]\n- [构建像人一样学习和思考的机器](notes\u002Fbuilding-machines-that-learn-and-think-like-people.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1604.00289)]\n- 基于阶梯网络的语言识别半监督方法 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1604.00317)]\n- [具有随机深度的深度网络](notes\u002Fstochastic-depth.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.09382)]\n- PHOCNet：用于手写文档中单词定位的深度卷积神经网络 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1604.00187)]\n\n\n#### 2016年3月\n\n- 通过强化学习获取外部证据来改进信息抽取 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1603.07954)]\n- 一种用于句法分析和句子理解的快速统一模型 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.06021)]\n- [用于代码生成的潜在预测网络](notes\u002Flatent-predictor-networks.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.06744)]\n- 注意、推断、重复：使用生成模型进行快速场景理解 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.08575)]\n- 循环批归一化 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.09025)]\n- 基于字符的注意力机制的神经语言校正 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.09727)]\n- [在序列到序列学习中融入复制机制](notes\u002Fcopynet.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.06393)]\n- 如何不要评估你的对话系统 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.08023)]\n- [循环神经网络的自适应计算时间](notes\u002Fact-rnn.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.08983)]\n- 深度学习中卷积运算指南 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.07285)]\n- 彩色图像着色 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.08983)]\n- 通过拼图游戏无监督地学习视觉表征 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.09246)]\n- 使用循环神经网络生成事实型问题：3000万条事实型问答语料库 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.06807)]\n- 基于人格特征的神经对话模型 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.06155)]\n- [用于神经机器翻译的无显式分词字符级解码器](notes\u002Fchar-level-decoder.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.06147)]\n- 从零开始的多任务跨语言序列标注 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.06270)]\n- 用于文本处理的神经变分推断 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06038)]\n- 不丢失记忆的循环丢弃 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.05118)]\n- 深度生成模型中的单次泛化 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.05106)]\n- 具有注意力建模的递归循环网络，用于野外光学字符识别 [[arXiv](Recursive Recurrent Nets with Attention Modeling for OCR in the Wild)]\n- 一种新的可视化深度神经网络的方法 [[arXiv](A New Method to Visualize Deep Neural Networks)]\n- 用于命名实体识别的神经架构 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.01360)]\n- 通过双向LSTM-CNN-CRF实现端到端序列标注 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.01354)]\n- 基于字符的神经机器翻译 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.00810)]\n- 学习词语分割表示以提升中文社交媒体的命名实体识别 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1603.00786)]\n\n#### 2016年2月\n\n- 循环神经网络的架构复杂度度量 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.08210)]\n- 权重归一化：一种简单的重新参数化方法，用于加速深度神经网络的训练 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.07868)]\n- 循环神经网络文法 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.07776)]\n- 视觉图谱：利用众包密集图像标注连接语言与视觉 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.07332)]\n- [用于大规模自然语言处理任务的上下文LSTM（CLSTM）模型](notes\u002Fclstm-large-scale.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.06291)]\n- 用于文本摘要的序列到序列RNN [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.06023)]\n- 从标注文档中提取重要句子 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1412.6815)]\n- 从无标签数据中学习句子的分布式表示 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.03483)]\n- 神经网络中深度的优势 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.04485)]\n- [联想长短时记忆网络](notes\u002Fassociative-lstm.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.03032)]\n- “我为什么要信任你？”：解释任何分类器的预测 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1602.04938)] [[代码](https:\u002F\u002Fgithub.com\u002Fmarcotcr\u002Flime)]\n- 利用循环对抗网络生成图像 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.05110)]\n- [探索语言模型的极限](notes\u002Fexploring-the-limits-of-lm.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.02410)]\n- Swivel：通过关注缺失的信息来改进词嵌入 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.02215)]\n- [WebNav：一项基于自然语言的新型大规模序列决策任务](notes\u002Fwebnav.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.02261)]\n- [结合卷积层和循环层实现高效的字符级文档分类](notes\u002Fefficient-char-level-document-classification-cnn-rnn.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.00367)]\n- 梯度下降收敛于极小值点 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1602.04915)] [[文章](http:\u002F\u002Fwww.offconvex.org\u002F2016\u002F03\u002F24\u002Fsaddles-again\u002F)]\n- BinaryNet：训练权重和激活值被约束为+1或-1的深度神经网络 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.02830)]\n- 通过标签一致的神经网络学习判别特征 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1602.01168)]\n\n#### 2016年1月\n\n- 你的机器学习测试得分是多少？面向机器学习生产系统的评分标准 [[Google研究](https:\u002F\u002Fresearch.google.com\u002Fpubs\u002Fpub45742.html)]\n- 像素循环神经网络 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.06759)]\n- 位运算神经网络 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.06071)]\n- 用于机器阅读的长短时记忆网络 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.06733)]\n- 基于覆盖率的神经机器翻译 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.04811)]\n- 理解深度卷积网络 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.04920)]\n- 通过扩散训练循环神经网络 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.04114)]\n- 图像自动描述生成：模型、数据集和评估指标综述 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.03896)]\n- [具有共享注意力机制的多路、多语种神经机器翻译](notes\u002Fmulti-way-nmt-shared-attention.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.01073)]\n- [用于语言建模的循环记忆网络](notes\u002Frmn-language-modeling.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.01272)]\n- 利用神经注意力机制将语言转换为逻辑形式 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.01280)]\n- 学习组合神经网络以进行问答 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.01705)]\n- 概率的必然性：在使用非概率反馈训练的通用神经网络中进行概率推理 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.03060)]\n- COCO-Text：自然图像中文本检测与识别的数据集及基准测试 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.07140)]\n- 基于注意力的RNN模型及其在计算机视觉中的应用综述 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1601.06823)]\n\n#### 2015年12月\n\n自然语言处理\n\n- [训练大词汇量神经语言模型的策略](notes\u002Fstrategies-for-training-large-vocab-lm.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1512.04906)]\n- [基于字节的多语言语言处理](notes\u002Fmultilingual-language-processing-from-bytes.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1512.00103)]\n- [通过预测N-gram学习文档嵌入，用于长篇电影评论的情感分类](notes\u002Flearning-document-embeddings-ngrams.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1512.08183)]\n- [基于长短时记忆网络的目标依赖情感分类](notes\u002Ftarget-dependent-sentiment-lstm.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1512.01100)]\n- 使用卷积神经网络在野外阅读文本 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1412.1842)]\n\n计算机视觉\n\n- [用于图像识别的深度残差学习](notes\u002Fdeep-residual-learning.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1512.03385)]\n- 重新思考用于计算机视觉的Inception架构 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1512.00567)]\n- 内外网：利用跳跃池化和循环神经网络在上下文中检测目标物体 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1512.04143)]\n- Deep Speech 2：英语和普通话的端到端语音识别 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1512.02595)]\n\n\n#### 2015年11月\n\n自然语言处理\n\n- [具有自然语言动作空间的深度强化学习](notes\u002Fdrl-nlp-action.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1511.04636)]\n- 基于循环神经网络的序列级训练 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06732)]\n- [教机器阅读和理解](notes\u002Fteaching-machines-to-read-and-comprehend.md) [[arxiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1506.03340)]\n- [半监督序列学习](notes\u002Fsemi-supervised-sequence-learning.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.01432)]\n- [多任务序列到序列学习](notes\u002Fmultitask-seq2seq.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06114)]\n- [字符级RNN的替代结构](notes\u002Falternative-structure-char-rnn.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06303)]\n- [更大上下文的语言建模](notes\u002Flarger-context-lm.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.03729)]\n- [统一标注方案：带有词嵌入的双向LSTM循环神经网络](notes\u002Funified-tagging-blstm.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.00215)]\n- 通往通用释义句子嵌入之路 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.08198)]\n- BlackOut：使用超大词汇表加速循环神经网络语言模型 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06909)]\n- 基于循环神经网络的序列级训练 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06732)]\n- 基于分布式表示的自然语言理解 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.07916)]\n- sense2vec - 神经网络词嵌入中快速准确的词义消歧方法 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06388)]\n- 基于LSTM的深度学习模型用于非事实型答案选择 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.04108)]\n\n程序\n\n- 神经随机存取机 [[arxiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06392)]\n- 神经程序员：通过梯度下降诱导潜在程序 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.04834)]\n- 神经程序员-解释器 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06279)]\n- 从示例中学习简单算法 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.07275)]\n- 神经GPU学习算法 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.08228)] [[代码](https:\u002F\u002Fgithub.com\u002Ftensorflow\u002Ftensor2tensor)]\n- 关于学习思考：强化学习控制器与循环神经网络世界模型新组合的算法信息论 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.09249)]\n\n视觉\n\n- ReSeg：用于目标分割的循环神经网络 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.07053)]\n- 解构阶梯网络架构 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06430)]\n- 使用深度卷积生成对抗网络进行无监督表征学习 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06434)]\n- 通过空洞卷积进行多尺度上下文聚合 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1511.07122)] [[代码](https:\u002F\u002Fgithub.com\u002Ffyu\u002Fdrn)]\n\n通用\n\n- 朝着原则性的无监督学习方向 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06440)]\n- 动态容量网络 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.07838)]\n- [从`ous空间生成句子](notes\u002Fgenerating-sentences-cont-space.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06349)]\n- Net2Net：通过知识迁移加速学习 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.05641)]\n- 通往机器智能的路线图 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.08130)]\n- 基于会话的推荐系统与循环神经网络 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.06939)]\n- 通过稳定激活来正则化RNN [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1511.08400)]\n\n\n#### 2015年10月\n\n- [卷积神经网络用于句子分类的敏感性分析（及从业者指南）](notes\u002Fsensitivity-analysis-cnn-sentence-classification.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1510.03820)]\n- [带有意图的注意力机制用于神经网络对话模型](notes\u002Fattention-with-intention.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1510.08565)]\n- 带有双向长短期记忆循环神经网络的词性标注 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1510.06168)]\n- 综述：深度学习领域的时光之旅——深度学习模型简介及其从最初想法发展至今的过程 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1510.04781)]\n- 自然语言处理中神经网络模型入门 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1510.00726)]\n- [促进多样性的神经对话模型目标函数](notes\u002Fdiversity-promoting-objective-ncm.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1510.03055)]\n\n\n#### 2015年9月\n\n- [用于文本分类的字符级卷积网络](notes\u002Fcharacter-level-cnn-for-text-classification.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1509.01626)]\n- [用于摘要式句子摘要的神经注意力模型](notes\u002Fneural-attention-model-for-abstractive-sentence-summarization.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1509.00685)]\n- 扑克-CNN：一种用于扑克游戏中下注和跟注的模式学习策略 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1509.06731)]\n\n#### 2015年8月\n\n- [基于子词单元的稀有词神经机器翻译](notes\u002Fnmt-subword.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1508.07909)] [[代码](https:\u002F\u002Fgithub.com\u002Frsennrich\u002Fsubword-nmt)]\n- 听、注意并拼写 [[arxiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1508.01211)]\n- [字符感知的神经语言模型](notes\u002Fcharacter-aware-nlm.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1508.06615)]\n- 通过LSTM对字符而非单词建模改进基于转换的句法分析 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1508.00657)]\n- 在形式中寻找功能：用于开放词汇词表示的组合式字符模型 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1508.02096)]\n- [基于注意力的神经机器翻译的有效方法](notes\u002Feffective-approaches-nmt-attention.md) [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1508.04025)]\n\n#### 2015年7月\n\n- [使用生成式分层神经网络模型构建端到端对话系统](e2e-dialog-ghnnm.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1507.04808)]\n- 基于阶梯网络的半监督学习 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1507.02672)]\n- [使用段落向量进行文档嵌入](notes\u002Fdocument-embedding-with-pv.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1507.07998)]\n- [训练非常深的网络](notes\u002Ftraining-very-deep-networks.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1507.06228)]\n\n#### 2015年6月\n\n- 丢弃法作为贝叶斯近似：在深度学习中表示模型不确定性 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1506.02142)]\n- [基于神经网络的上下文敏感会话响应生成](notes\u002Fnn-context-sentitive-responses.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1506.06714)]\n- [使用段落向量进行文档嵌入](notes\u002Fdocument-embedding-with-pv.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1507.07998)]\n- [神经对话模型](notes\u002Fneural-conversational-model.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1506.05869)]\n- [跳字向量](notes\u002Fskip-thought-vectors.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1506.06726)]\n- [指针网络](notes\u002Fpointer-networks.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1506.03134)]\n- [空间变换网络](notes\u002Fspatial-transformer-networks.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1506.02025)]\n- 在无树结构架构的神经网络中实现树状组合 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1506.04834)]\n- 可视化与理解自然语言处理中的神经模型 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1506.01066)]\n- 学习具有无界记忆的转导 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1506.02516)]\n- 随便问我：用于自然语言处理的动态记忆网络 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1506.07285)]\n- [深度知识追踪](notes\u002Fdeep-knowledge-tracing.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1506.05908)]\n\n#### 2015年5月\n\n- [ReNet：一种基于循环神经网络的卷积网络替代方案](notes\u002Frenet-rnn-alternative-to-convnet.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1505.00393)]\n- 强化学习神经图灵机 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1505.00521)]\n\n#### 2015年4月\n\n- 相关性神经网络 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1504.07225)]\n\n#### 2015年3月\n\n\n- [蒸馏神经网络中的知识](notes\u002Fdistilling-the-knowledge-in-a-nn.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1503.02531)]\n- [端到端记忆网络](notes\u002Fend-to-end-memory-networks.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1503.08895)]\n- [用于短文本对话的神经应答机](notes\u002Fneural-responding-machine.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1503.02364)]\n- [批量归一化：通过减少内部协变量偏移加速深度网络训练](notes\u002Fbatch-normalization.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1502.03167)]\n- 摆脱鞍点——张量分解的在线随机梯度下降 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1503.02101)] [[文章](Escaping from Saddle Points)]\n\n\n#### 2015年2月\n\n- 通过深度强化学习实现人类水平控制 [[Nature](https:\u002F\u002Fweb.stanford.edu\u002Fclass\u002Fpsych209\u002FReadings\u002FMnihEtAlHassibis15NatureControlDeepRL.pdf)] [[代码](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Fdqn)]\n- [从零开始理解文本](notes\u002Ftext-understanding-from-scratch.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1502.01710)]\n- [展示、注意与讲述：结合视觉注意力的神经图像字幕生成](notes\u002Fshow-attend-tell.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1502.03044)]\n\n#### 2015年1月\n\n- 机器学习系统中的隐性技术债务 [[NIPS](https:\u002F\u002Fpapers.nips.cc\u002Fpaper\u002F5656-hidden-technical-debt-in-machine-learning-systems.pdf)]\n\n#### 2014年12月\n\n- 在循环神经网络中学习更长的记忆 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1412.7753)]\n- [神经图灵机](notes\u002Fneural-turing-machines.md) [[arxiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1410.5401)]\n- [语法作为外语](notes\u002Fgrammar-as-a-foreign-language.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1412.7449)]\n- [关于在神经机器翻译中使用超大目标词汇表](notes\u002Fon-using-very-large-target-vocabulary-for-nmt.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1412.2007)]\n- 卷积神经网络中有效利用词序进行文本分类 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1412.1058v1)]\n- 基于视觉注意力的多对象识别 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1412.7755)]\n\n#### 2014年11月\n\n- 多层网络的损失曲面 [[arXiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F1412.0233)]\n\n#### 2014年10月\n\n- [学习执行](notes\u002Flearning-to-execute.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1410.4615)]\n\n#### 2014年9月\n\n- [使用神经网络进行序列到序列学习](notes\u002Fseq2seq-with-neural-networks.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1409.3215)]\n- [通过联合学习对齐与翻译实现神经机器翻译](notes\u002Fnmt-jointly-learning-to-align-and-translate.md) [[arxiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1409.0473)]\n- [关于神经机器翻译的性质：编码器-解码器方法](notes\u002Fproperties-of-neural-mt.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1409.1259)]\n- [循环神经网络正则化](notes\u002Frnn-regularization.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1409.2329)]\n- 用于大规模图像识别的超深卷积神经网络 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1409.1556)]\n- 使用卷积进一步深入 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1409.4842)]\n\n#### 2014年8月\n\n- 用于句子分类的卷积神经网络 [[arxiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1408.5882)]\n\n#### 2014年7月\n\n#### 2014年6月\n\n- [使用RNN编码器-解码器学习短语表示以进行统计机器翻译](notes\u002Flearning-phrase-representations.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1406.1078)]\n- [视觉注意力的循环模型](notes\u002Frecurrent-models-of-visual-attention.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1406.6247)]\n- 生成对抗网络 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1406.2661)]\n\n#### 2014年5月\n\n- [句子和文档的分布式表示](notes\u002Fdistributed-representations-of-sentences-and-documents.md) [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1405.4053)]\n\n#### 2014年4月\n\n- 用于建模句子的卷积神经网络 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1404.2188)]\n\n#### 2014年3月\n\n#### 2014年2月\n\n#### 2014年1月\n\n- 机器学习：技术债务的高息信用卡 [[Google Research](https:\u002F\u002Fresearch.google.com\u002Fpubs\u002Fpub43146.html)]\n\n#### 2013年\n\n- 可视化与理解卷积网络 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1311.2901)]\n- DeViSE：一种深度视觉-语义嵌入模型 [[出版物](http:\u002F\u002Fresearch.google.com\u002Fpubs\u002Fpub41473.html)]\n- Maxout网络 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1302.4389)]\n- 利用语言间的相似性进行机器翻译 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1309.4168)]\n- 向量空间中单词表示的有效估计 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1301.3781)]\n\n\n#### 2011年\n\n- 自然语言处理（几乎）从零开始 [[arXiv](http:\u002F\u002Farxiv.org\u002Fabs\u002F1103.0398)]","# deeplearning-papernotes 快速上手指南\n\n`deeplearning-papernotes` 并非一个需要编译或运行的软件库，而是一个**深度学习论文笔记与资源索引集合**。它按时间顺序整理了重要的 AI 研究论文、相关技术文章及开源代码链接。本指南将帮助你快速获取并利用这些资源。\n\n## 环境准备\n\n由于本项目主要为文档索引，无复杂的系统依赖，仅需基础的网络环境和阅读工具。\n\n*   **操作系统**：Windows \u002F macOS \u002F Linux 均可。\n*   **前置依赖**：\n    *   现代浏览器（推荐 Chrome, Edge 或 Firefox）用于访问论文链接。\n    *   Git（可选，用于克隆仓库到本地离线浏览）。\n    *   PDF 阅读器（用于查看 arXiv 论文）。\n*   **网络建议**：\n    *   部分链接指向 arXiv、Google DeepMind 或 GitHub，国内访问可能不稳定。\n    *   **推荐方案**：使用学术加速镜像访问 arXiv 论文。例如，将链接中的 `arxiv.org` 替换为 `arxiv.org.cn` 或使用国内高校镜像站。\n\n## 安装步骤\n\n你可以通过以下两种方式获取资源：\n\n### 方式一：在线浏览（推荐）\n直接访问 GitHub 仓库页面查看整理好的列表：\n```bash\n# 在浏览器中打开\nhttps:\u002F\u002Fgithub.com\u002Fterryum\u002Fawesome-deep-learning-papers (注：此处为示例逻辑，实际请访问原项目地址)\n```\n*注意：原项目通常为静态 Markdown 文件，直接在 GitHub 网页阅读体验最佳。*\n\n### 方式二：克隆到本地\n如果你希望离线查阅或自行整理，可克隆仓库：\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Fterryum\u002Fdeeplearning-papernotes.git\ncd deeplearning-papernotes\n```\n\n## 基本使用\n\n本项目的使用核心在于**按需检索**和**顺藤摸瓜**获取代码与数据。\n\n### 1. 查找目标论文\n打开项目根目录下的 `README.md` 文件（或在 GitHub 页面），利用浏览器的搜索功能（`Ctrl+F` 或 `Cmd+F`）查找关键词。\n\n*   **示例场景**：你想寻找关于 \"GAN\" (生成对抗网络) 的最新进展。\n*   **操作**：搜索关键词 `GAN`。\n*   **结果定位**：你会在 `2017-11` 或 `2017-10` 月份列表中找到如下条目：\n    > - Progressive Growing of GANs for Improved Quality, Stability, and Variation [[Research at Nvidia](...)] [[code](https:\u002F\u002Fgithub.com\u002Ftkarras\u002Fprogressive_growing_of_gans)]\n\n### 2. 获取代码与复现\n大多数条目都附带了 `[code]` 标签，直接点击即可跳转至对应的 GitHub 开源项目。\n\n*   **操作步骤**：\n    1.  点击条目后的 `[[code](URL)]` 链接。\n    2.  进入对应仓库后，参照该独立仓库的 `README` 进行环境配置和运行。\n    \n    *以 StarGAN 为例：*\n    ```bash\n    # 1. 从笔记中找到 StarGAN 的代码链接并克隆\n    git clone https:\u002F\u002Fgithub.com\u002Fyunjey\u002FStarGAN.git\n    cd StarGAN\n\n    # 2. 按照该仓库的具体要求安装依赖 (示例)\n    pip install -r requirements.txt\n\n    # 3. 运行训练或测试脚本\n    python main.py --mode train --dataset CelebA\n    ```\n\n### 3. 阅读深度解析\n对于理论性较强的论文（如 *The Matrix Calculus You Need For Deep Learning*），建议优先点击 `[[article]]` 链接（如果有），通常这些链接指向官方博客或技术解读文章，比原始论文更易于理解。\n\n---\n**提示**：该列表涵盖了从 2017 年至 2018 年初的经典工作（如 AlphaGo Zero, Transformer 变体，DeepType 等），是回顾深度学习发展史和寻找经典基线模型（Baseline）的绝佳起点。","某自动驾驶初创公司的算法工程师正急需为无人机编队开发一套能在复杂野外环境中稳定飞行的视觉导航系统，同时需要快速复现前沿的强化学习策略。\n\n### 没有 deeplearning-papernotes 时\n- **文献检索如大海捞针**：面对 arXiv 上每天涌现的数百篇论文，难以快速筛选出与“无人机飞行（DroNet）”或“分布式强化学习（IMPALA）”直接相关的核心文章，大量时间浪费在浏览标题和摘要上。\n- **数学推导门槛高**：在阅读《深度学习所需的矩阵微积分》等基础理论论文时，复杂的公式推导往往让工程师卡壳数天，难以将理论转化为代码逻辑。\n- **复现路径不清晰**：即使找到了像 DensePose 或 Soft Actor-Critic 这样的关键论文，也常因找不到官方代码链接、配套文章或具体的实验设置细节，导致复现工作反复试错，进度严重滞后。\n- **知识体系碎片化**：团队内部缺乏对“在线学习”或“可解释性”等综述类文章的统一认知，导致技术选型时各自为战，难以形成合力。\n\n### 使用 deeplearning-papernotes 后\n- **精准定位前沿成果**：通过按月整理的清单，工程师能瞬间锁定 2018 年初发布的 DroNet 和 IMPALA 论文，并直接获取对应的 GitHub 代码库和 DeepMind 官方博客解读，将调研时间从数周缩短至几小时。\n- **攻克理论难点**：借助工具中收录的矩阵微积分笔记和深度批判性评估文章，团队快速补齐了数学短板，顺利推导出适合无人机传感器数据的损失函数。\n- **一站式复现资源**：每个条目都附带了论文、技术文章和源代码的直接链接，工程师可以直接基于 SBNet 或 DeepType 的开源实现进行微调，大幅降低了环境配置和调试成本。\n- **构建系统化认知**：利用工具提供的综述类笔记（如视觉可解释性调查），团队迅速统一了技术路线图，明确了从感知到决策的最优架构组合。\n\ndeeplearning-papernotes 通过将分散的顶会论文、代码实现与通俗解读结构化整合，成为了深度学习研究者从理论通往工程落地的最高效桥梁。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fdennybritz_deeplearning-papernotes_ebc3e7c4.png","dennybritz","Denny Britz","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fdennybritz_a0c7fe5c.jpg",null,"Tokyo, Japan","https:\u002F\u002Fdennybritz.com","https:\u002F\u002Fgithub.com\u002Fdennybritz",4418,900,"2026-04-13T09:54:04","","未说明",{"notes":89,"python":87,"dependencies":90},"该仓库（deeplearning-papernotes）并非一个可运行的 AI 软件工具，而是一个深度学习论文笔记和资源的整理列表。README 内容仅包含按月份分类的论文标题、arXiv 链接、相关文章及对应代码仓库的外部链接。因此，该项目本身没有操作系统、GPU、内存、Python 版本或依赖库的安装需求。用户若需运行列表中提及的具体算法，需分别访问各论文对应的独立代码仓库查看其特定环境要求。",[],[18],"2026-03-27T02:49:30.150509","2026-04-13T23:54:43.513324",[],[]]