ICLR 2020 高质量强化学习论文汇总

2019 年 11 月 11 日 极市平台

加入极市专业CV交流群，与6000+来自腾讯，华为，百度，北大，清华，中科院等名企名校视觉开发者互动交流！更有机会与李开复老师等大牛群内互动！

同时提供每月大咖直播分享、真实项目需求对接、干货资讯汇总，行业技术交流。关注 极市平台 公众号，回复 加群，立刻申请入群~

作者：Chersophyte

来源：https://zhuanlan.zhihu.com/p/90393651

本文已经作者授权，转载请联系原作者

简单做个今年ICLR上我个人觉得质量比较高的RL方向的工作总结。

有很多疏漏，漏了没看到的工作后续再补吧。

质量比较高的意思不是说都是高分工作，有的分数差异比较大的我也放上来了。

加了一些tranfer reinforcement learning有关的工作。

Adaptation:

CM3: Cooperative Multi-goal Multi-stage Multi-agent Reinforcement Learning

链接：
https://openreview.net/forum?id=S1lEX04tPr&noteId=r1e0fqAaKr

Adapt-to-Learn: Policy Transfer in Reinforcement Learning

链接：
https://openreview.net/forum?id=ryeT10VKDH&noteId=S1lVR3-bqS

Evolutionary Population Curriculum for Scaling Multi-Agent Reinforcement Learning

链接：
https://openreview.net/forum?id=SJxbHkrKDH&noteId=ryxYvoLRYH

MULTIPOLAR: Multi-Source Policy Aggregation for Transfer Reinforcement Learning between Diverse Environmental Dynamics

链接：https://openreview.net/forum?id=Byx9p2EtDH

Sparse-reward:

Intrinsic Motivation for Encouraging Synergistic Behavior

链接:
https://openreview.net/forum?id=SJleNCNtDH&noteId=SJxFBZDpYr

Two-player zero-sum extensive-games with imperfect information (TZIEG):

Posterior sampling for multi-agent reinforcement learning: solving extensive games with imperfect information

链接：
https://openreview.net/forum?id=Syg-ET4FPS&noteId=BkgacWb0cS

Human Interaction:

Simplified Action Decoder for Deep Multi-Agent Reinforcement Learning

链接：
https://openreview.net/forum?id=B1xm3RVtwB&noteId=Bkl2ONldFr

Communication:

Graph Convolutional Reinforcement Learning

链接：
https://openreview.net/forum?id=HkxdQkSYDB&noteId=SygTyGcOcr

Learning Nearly Decomposable Value Functions Via Communication Minimization

链接：
https://openreview.net/forum?id=HJx-3grYDB&noteId=SygnB7pe5S

Multi-agent Reinforcement Learning for Networked System Control

链接：https://openreview.net/forum?id=Syx7A3NFvH

Learning Structured Communication for Multi-agent Reinforcement Learning

链接：https://openreview.net/forum?id=BklWt24tvH

Interaction Modelling:

Promoting Coordination through Policy Regularization in Multi-Agent Deep Reinforcement Learning

链接：https://openreview.net/forum?id=BkggGREKvS

Training, Exploration:

A Generalized Training Approach for Multiagent Learning

链接：
https://openreview.net/forum?id=Bkl5kxrKDr&noteId=r1xBh9CaYS

Influence-Based Multi-Agent Exploration

链接：
https://openreview.net/forum?id=BJgy96EYvr&noteId=HJlVKGuwdH

Learning Expensive Coordination: An Event-Based Deep RL Approach

链接：https://openreview.net/forum?id=ryeG924twB

Imitation, Inverse:

Asynchronous Multi-Agent Generative Adversarial Imitation Learning

链接：
https://openreview.net/forum?id=Syx33erYwH&noteId=r1l9cYFddS

Multi-Agent Interactions Modeling with Correlated Policies

链接：https://openreview.net/forum?id=B1gZV1HYvS

-End-

*延伸阅读

CV细分方向交流群

添加极市小助手微信（ID : cv-mart），备注：研究方向-姓名-学校/公司-城市（如：目标检测-小极-北大-深圳），即可申请加入目标检测、目标跟踪、人脸、工业检测、医学影像、三维&SLAM、图像分割等极市技术交流群（已经添加小助手的好友直接私信），更有每月大咖直播分享、真实项目需求对接、干货资讯汇总，行业技术交流，一起来让思想之光照的更远吧~