最近机器学习顶会ICML2021接收结果已经公布，共有5513篇论文投稿，共有1184篇接受（包括1018篇短论文和166篇长论文），接受率21.48%。 专知在这里整理来自Twitter、arXiv、知乎放出来的20篇最新ICML论文，方便大家抢先阅览！这些论文包括图神经网络、因果性、强化学习等等。
1. 等变图神经网络，E(n) Equivariant Graph Neural Networks
https://www.zhuanzhi.ai/paper/367751014c431bb755d1f9ab5ee31357
2. 图神经网络优化，Optimization of Graph Neural Networks: Implicit Acceleration by Skip Connections and More Depth
https://www.zhuanzhi.ai/paper/609aef10a18ac8eb66f1d1873c8ec445
3. 图神经网络加速训练，GraphNorm: A Principled Approach to Accelerating Graph Neural Network Training
https://www.zhuanzhi.ai/paper/e80615a1e4e175a6b2e80c486a3d3baa
4. 图对抗网络，Information Obfuscation of Graph Neural Networks
https://www.zhuanzhi.ai/paper/a0e171fd79f56fa03ac6f69bf1135c18
5. 因果表示学习，Causal Curiosity: RL Agents Discovering Self-supervised Experiments for Causal Representation Learning
https://www.zhuanzhi.ai/paper/2e260f8ebc9a06274446b41a6a692564
6.元强化学习，Exploration in Approximate Hyper-State Space for Meta Reinforcement Learning
https://www.zhuanzhi.ai/paper/f7f81c8eba46da07eb60f30479055207
7. 联邦持续学习，Federated Continual Learning with Weighted Inter-client Transfer
https://www.zhuanzhi.ai/paper/dd39c0ce7db9b80aed29755c9bff2ea9
8. 带不确定性的领域不变学习，A Bit More Bayesian: Domain-Invariant Learning with Uncertainty
https://arxiv.org/abs/2105.04030
9. 领域自适应标签传播理论Theory of Label Propagation for Domain Adaptation
https://www.zhuanzhi.ai/paper/ba2fbf77ca95d4727e07dc73f5c4c9ec
10. MC-LSTM: Mass-Conserving LSTM
https://www.zhuanzhi.ai/paper/3ebaa454e1fe52de110c8f560e6c15c3
11. 特征属性形式化，Towards Rigorous Interpretations: a Formalisation of Feature Attribution
https://www.zhuanzhi.ai/paper/b103d375a6903682211587307e3cf183
12. 消息传递简单网络，Weisfeiler and Lehman Go Topological: Message Passing Simplicial Networks
https://www.zhuanzhi.ai/paper/2715944f774deafd4979f10730a47fd2
13.
剖监督对比学习，Dissecting Supervised Constrastive Learning
https://www.zhuanzhi.ai/paper/e223c4dde4ab3911bf4ae98da7d3665a
14. 不平衡小批量最优传输;域自适应的应用，Unbalanced minibatch Optimal Transport; applications to Domain Adaptation
https://www.zhuanzhi.ai/paper/d345a2b0e86bafa301144f733b24aac4
15. 复杂行动空间中的学习与规划，Learning and Planning in Complex Action Spaces
https://www.zhuanzhi.ai/paper/dd21770f6c7ccb474fc3cb2d4817d476
16. 神经A*搜索的路径规划，Path Planning using Neural A* Search
https://www.zhuanzhi.ai/paper/670e8c6cd4c9cbd19ca07677ab9d1df5
17. 有向偏差放大，Directional Bias Amplification
https://www.zhuanzhi.ai/paper/8252bd00288b1d8f42b0f22eb48b8497
18. 现代强化学习，Revisiting Peng’s Q(λ) for Modern Reinforcement Learning
https://arxiv.org/abs/1912.13200.
19. 池化注意力的长文档建模，Poolingformer: Long Document Modeling with Pooling Attention
https://www.zhuanzhi.ai/paper/3d04de5c54e6026e7a6090e9b64017d3
20.
分子图神经网络可解释性，Improving Molecular Graph Neural Network Explainability
with Orthonormalization and Induced Sparsity
https://www.zhuanzhi.ai/paper/47815c0b1748057eae88b9109f29a5c9
专知便捷查看
便捷下载，请关注专知公众号（点击上方蓝色专知关注）
专知，专业可信的人工智能知识分发，让认知协作更快更好！欢迎注册登录专知www.zhuanzhi.ai，获取5000+AI主题干货知识资料！欢迎微信扫一扫加入专知人工智能知识星球群，获取最新AI专业干货知识教程资料和与专家交流咨询！
点击“阅读原文”，了解使用专知，查看获取5000+AI主题知识资源
