MADRL

다중에이전트 강화학습

정보 : MADRL 슬로우 페이퍼

참가자 : 강효림, 김예찬, 노원종, 박규봉, 배영민, 안홍일, 양홍선, 정재현

퍼실 : 김예찬

논문리스트

2015

• (Review) Multiagent Cooperation and Competition with Deep Reinforcement Learning(1511)
• (Review) Deep Reinforcement Learning in Parameterized Action Space(1511)

2016

• (Review) Learning to Communicate to Solve Riddles with Deep distributed Recurrent Q-Network(1602)
• (Review) Deep Reinforcement Learning from Self-Play in Imperfect-Information Games(1603)
• Opponent Modeling in Deep Reinforcement Learning(1609)
• (Review) Learning to Communicate with Deep Multi-Agent Reinforcement Learning(1605)
• (Review) Learning Multiagent Communication with Backpropagation(1605)
• Safe, Multi-Agent, Reinforcement Learning for Autonomous Driving(1610)
• (Review) Learning to Play Guess Who? and Inventing a Grounded Language as a Consequence(1611)
• (Review) Multi-Agent Cooperation and the Emergence of (Natural) Language(1612)

2016 참고

• Interaction Networks for Learning about Objects, Relations and Physics(1612)

2017

• (Review) Cooperative Multi-Agent Control using Deep Reinforcement Learning
• (Review) Multi-agent Reinforcement Learning in Sequential Social Dilemmas(1702)
• (Review) Stabilising Experience Replay for Deep Multi-Agent Reinforcement Learning(1702)
• Coordinated Multi-Agent Imitation Learning(1703)
• (Review) Emergence of Grounded Language in Multi-agent Populations(1703)
• Deep Decentralized Multi-task Multi-Agent Reinforcement Learning under Parital Observability(1703)
• (Review) Counterfactual Multi-Agent Policy Gradients(1705)
• Multiagent Bidirectionally-Coordinated Nets Emergence of Human-level Coordination in Learning to Play StarCraft Combat Games(1703)
• (Review) Emergence of Language with Multi-agent Games: Learning to Communicate with Sequence of Symbols(1705)
• (Review) Multi-Agent Actor-Critic for Mixed Cooperative-Competitive Environments(1706)
• Value-Decomposition Networks for Cooperative Multi-Agent Learning(1706)
• (Review) VAIN: Attentional Multi-agent Predictive Modeling(1706)
• (Review) Learning with Opponent-Learning Awareness(1709)
• (Review) Emergence Complexity via Multi-Agent Competition(1710)
• Continuous Adaptation via Meta-Learning in Nonstationary and Competitive Environments(1710)
• (Review) A Unified Game-Theoretic Approach to Multiagent Reinforcement Learning(1711)

2017 참고

• DeepStack: Expert-Level Artificial Intelligence in No-Limit Poker(1701)
• Robust Adversarial Reinforcement Learning(1703)
• Intrinsic Motivation and Automatic Curricula via Asymmetric Self-Play(1703)

2018

• DiCE: The Infinitely Differentiable Monte-Carlo Estimator(1802)
• Fully Decentralized Multi-Agent Reinforcement Learning with Networked Agents
• Modeling Others using Oneself in Multi-Agent Reinforcement Learning(1802)
• (Review) QMIX: Monotonic Value Function Factorisation for Deep Multi-Agent Reinforcement Learning(1803)
• (Review) Emergence of Linguistic Communication from Referential Games with Symbolic and Pixel Input(1804)
• (Review) Emergent Communication through Negotiation(1804)
• (Review) Learning Attentional Communication for Multi-Agent Cooperation(1805)
• Learning to Teach in Cooperative Multiagent Reinforcement Learning(1805)
• (Review) Learning Policy Representation in Multiagent Systems(1805)
• Human-level performance in first-person multiplayer games with population-based deep reinforcement learning(1807)
• Learning to Coordinate with Coordination Graphs in Repeated Single-State Multi-Agent Decision Problems
• Learning to act in Decentralized Partially Observable MDPs
• Relational Forward Models for Multi-Agent Learning(1809)
• M^3RL: Mind-aware Multi-agent Management Reinforcement Learning(1810)
• (Review) TarMAC: Targeted Multi-Agent Communication
• Bayesian Action Decoder for Deep Multi-Agent Reinforcement Learning(1811)
• (Review)Deep Multi-Agent Reinforcement Learning with Relevance Graphs(1811)
• Multi-agent Deep Reinforcement Learning with Extremely Noisy Observations(1812)
• Learning when to Communicate at Scale in Multiagent Cooperative and Competitive Tasks(1812)

2019

• Improving Coordination in Multi-Agent Deep Reinforcement Learning through Memory-driven Communication(1901)
• Probabilistic Recursive Reasoning for Multi-Agent Reinforcement Learning(1901)
• Learning to Schedule Communication in Multi-agent Reinforcement Learning(1902)
• (Review) Message-Dropout: An Efficient Training Method for Multi-Agent Reinforcement Learning(1902)
• Emergent Coordination Through Competition(1902)
• Robust Multi-Agent Reinforcement Learning via Minimax Deep Deterministic Policy Gradients
• ...

Schedule

Week01

• 자기소개 및 발표자 지정
• Intro

Week02

• Multiagent Cooperation and Competition with Depp Reinforcement Learning

  • Presenter : 김예찬
  • Paper : https://arxiv.org/abs/1511.08779
  • Material : 1. Multiagent Cooperation and Competition with Depp Reinforcement Learning

• Learning to Communicate to Solve Riddles with Deep distributed Recurrent Q-Network

  • Presenter : 김예찬
  • Paper : https://arxiv.org/abs/1602.02672
  • Material : 2. Learning to Communicate to Solve Riddles with Deep distributed Recurrent Q-Network

Week03

• Deep Reinforcement Learning from Self-Play in Imperfect-Information

  • Presenter : 강효림
  • Paper : https://arxiv.org/abs/1603.01121
  • Material : 3. Deep Reinforcement Learning from Self-Play in Imperfect-Information

• Learning to Communicate with Deep Multi-Agent Reinforcement Learning

  • Presenter : 강효림
  • Paper : https://arxiv.org/abs/1605.06676
  • Material : 4. Learning to Communicate with Deep Multi-Agent Reinforcement Learning

Week04

• Learning Multiagent Communication with Backpropagation

  • Presenter : 안홍일
  • Paper : https://arxiv.org/abs/1605.07736
  • Material : 5. Learning Multiagent Communication with Backpropagation

• Cooperative Multi-Agent Control using Deep Reinforcement Learning

  • Presenter : 배영민
  • Paper : http://ala2017.it.nuigalway.ie/papers/ALA2017_Gupta.pdf
  • Material : 6. Cooperative Multi-Agent Control using Deep Reinforcement Learning

Week05

• Stabilising Experience Replay for Deep Multi-Agent Reinforcement Learning

  • Presenter : 박규봉
  • Paper :
  • Material :

• Multi-agent Reinforcement Learning in Sequencial Social Dilemma

  • Presenter : 김예찬
  • Paper :
  • Material :

• Learning to Play Guess Who? and Inventing a Grounded Language as a Consequence

  • 사전학습 및 토론

Week06

• Emergence of Grounded Language in Multi-Agent Populations Games

  • Presenter : 양홍선
  • Paper :
  • Material :

• Emergence of Language with Multi-agent Games: Learning to Communicate with Sequence of Symbols

  • Presenter : 김예찬
  • Paper :
  • Material :

• Multi-Agent Cooperation and the Emergence of (Natural) Language(1612)

  • 사전학습 및 토론

Week07

• Counterfactual Multi-Agent Policy Gradients

  • Presenter : 김예찬
  • Paper :
  • Material :

• Multi-Agent Actor-Critic for Mixed Cooperative-Competitive Environments

  • Presenter : 김예찬
  • Paper :
  • Material :

Week08

• Learning with Opponent-Learning Awareness

  • Presenter : 박규봉
  • Paper :
  • Material :

• Emergence Complexity via Multi-Agent Competition

  • Presenter : 김예찬
  • Paper :
  • Material :

• VAIN: Attentional Multi-agent Predictive Modeling(1706)

  • 사전학습 및 토론

Week09

• A Unified Game-Theoretic Approach to Multiagent Reinforcement Learning

  • Presenter : 강효림
  • Paper :
  • Material :

• QMIX: Monotonic Value Function Factorisation for Deep Multi-Agent Reinforcement Learning

  • Presenter : 양홍선
  • Paper :
  • Material :

• Learning Attentional Communication for Multi-Agent Cooperation

  • 사전학습 및 토론

Week10

• Emergence of Linguistic Communication from Referential Games with Symbolic and Pixel Input

  • Presenter : 김예찬
  • Paper :
  • Material :

• Emergence Communication through Negotiation

  • Presenter : 김예찬
  • Paper :
  • Material :

Week11

• Learning Policy Representation in Multiagent Systems

  • Presenter : 박규봉
  • Paper :
  • Material :

• TarMAC: Targeted Multi-Agent Communication

  • Presenter : 김예찬
  • Paper :
  • Material :