雷锋网 AI 科技评论按：机器学习领域顶级会议 ICML 2017 已经开始了，雷锋网(公众号：雷锋网)记者会带来全方位的大会报道。
在之前的文章中，雷锋网 AI 科技评论就介绍过434篇 ICML 收录论文中有多达44篇都出现了谷歌的名字，谷歌的在机器学习领域的投入与成果之多可见一斑。今天谷歌也正式给出了自己的收录论文名单，署名的谷歌的就有42篇，其中有4篇是在几个 workshop 中。根据我们前两天的报道，署名DeepMind的收录论文也有25篇之多。那么来自谷歌的全部论文就有65篇（其中2篇是谷歌和DeepMind合作完成的），大约是 ICML 2017 全部收录论文的七分之一。这个数字简直大到让人有点害怕了。
谷歌在文中说，机器学习是谷歌的重点战略之一，他们有非常活跃的研究小组在领域内的各个方面进行研究，包括深度学习和更多的传统算法，理论和应用探索并重。谷歌的研究人员们运用可拓展的工具和架构，构建出各种各样的机器学习系统供他们解决语言、语音、翻译、音乐、视觉处理等等方面艰深的科学和工程问题。
作为机器学习领域的带头人之一，谷歌不仅是今年 ICML 2017的白金赞助商，也实实在在做出了许多研究成果（体现为42篇接收论文），此次参加会议展示论文、组织workshop的研究人员也有130人之多，热切地希望跟整个机器学习大家庭有更多的沟通和协作。
除了论文和workshop，谷歌的研究人员们还会对一些新的研究成果做讲解和展示，比如介绍 Facets 背后的技术、音频生成神经网络 Nsynth，还会有一个关于谷歌大脑培训生计划的问答活动。
谷歌在文中给出了自己的42篇论文列表，感兴趣的读者可以具体关注一下，打包下载地址见文末
- A Unified Maximum Likelihood Approach for Estimating Symmetric Properties of Discrete Distributions
- Accelerating Eulerian Fluid Simulation With Convolutional Networks
- AdaNet: Adaptive Structural Learning of Artificial Neural Networks
- Adaptive Feature Selection: Computationally Efficient Online Sparse Linear Regression under RIP
- Algorithms for ℓp Low-Rank Approximation
- Axiomatic Attribution for Deep Networks
- Bridging the Gap Between Value and Policy Based Reinforcement Learning
- Lifelong Learning: A Reinforcement Learning Approach Workshop论文，workshop时间8月10日
- Canopy Fast Sampling with Cover Trees
- Conditional Image Synthesis with Auxiliary Classifier GANs
- Consistent k-Clustering
- Deep Value Networks Learn to Evaluate and Iteratively Refine Structured Outputs
- Density Level Set Estimation on Manifolds with DBSCAN
- Device Placement Optimization with Reinforcement Learning
- Differentiable Programs with Neural Libraries
- Distributed Mean Estimation with Limited Communication
- Filtering Variational Objectives
- Deep Structured Prediction Workshop论文，workshop时间8月11日
- Generating High-Quality and Informative Conversation Responses with Sequence-to-Sequence Models
- Learning to Generate Natural Language Workshop论文，workshop时间8月10日
- Geometry of Neural Network Loss Surfaces via Random Matrix Theory
- Gradient Boosted Decision Trees for High Dimensional Sparse Output
- Input Switched Affine Networks: An RNN Architecture Designed for Interpretability
- Large-Scale Evolution of Image Classifiers
- Latent LSTM Allocation: Joint Clustering and Non-Linear Dynamic Modeling of Sequence Data
- Learned Optimizers that Scale and Generalize
- Learning Deep Latent Gaussian Models with Markov Chain Monte Carlo
- Learning to Generate Long-term Future via Hierarchical Prediction
- Maximum Selection and Ranking under Noisy Comparisons
- Neural Audio Synthesis of Musical Notes with WaveNet Autoencoders
- 谷歌与DeepMind合作论文
- Neural Message Passing for Quantum Chemistry
- 谷歌与DeepMind合作论文
- Neural Optimizer Search with Reinforcement Learning
- On the Expressive Power of Deep Neural Networks
- Online and Linear-Time Attention by Enforcing Monotonic Alignments
- Probabilistic Submodular Maximization in Sub-Linear Time
- REBAR: Low-variance unbiased gradient estimates for discrete latent variable models
- Deep Structured Prediction Workshop论文，workshop时间8月11日
- Robust Adversarial Reinforcement Learning
- RobustFill: Neural Program Learning under Noisy IO
- Sequence Tutor: Conservative Fine-Tuning of Sequence Generation Models with KL-control
- Sharp Minima Can Generalize For Deep Nets
- Stochastic Generative Hashing
- Tight Bounds for Approximate Carathéodory and Beyond
- Uniform Convergence Rates for Kernel Density Estimation
- Variational Boosting: Iteratively Refining Posterior Approximations
- Zero-Shot Task Generalization with Multi-Task Deep Reinforcement Learning
42篇谷歌署名论文+17篇DeepMind署名演讲论文打包下载链接:
http://pan.baidu.com/s/1jIFYZqu 密码: t74m
雷锋网 AI 科技评论记者也已经在 ICML现场参与大会活动，更多报道请继续关注。
相关文章：
太热了，雷锋网跑来悉尼避暑，顺便为你报道 ICML 2017
重磅 | ICML 2017最佳论文公布！机器学习的可解释性成热点
本文作者：杨晓凡
本文转自雷锋网禁止二次转载，原文链接