Cooperation and communication in multiagent deep reinforcement learning

By Matthew J. Hausknecht. 2016

38 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Reinforcement Learning, Game & Decision Theory

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Reinforcement Learning

In Game & Decision Theory

Citations in the corpus

Value-Decomposition Networks For Cooperative Multi-Agent Learning (2017, background): “Further, in the context of the introduced agent, we evaluate weight sharing, role information and information channels as additional enhancements that have recently been reported to improve sample complexity and memory requirements (Hausknecht, 2016; Foerster et al., 2016; Sukhbaatar et al., 2016).”
A Unified Game-Theoretic Approach to Multiagent Reinforcement Learning (2017, background): “In MARL, several agents interact and learn in an environment simultaneously, either competitively such as in Go [91] and Poker [39, 105, 74], cooperatively such as when learning to communicate [23, 93, 36], or some mix of the two [60, 95, 35].”
Emergent Coordination Through Competition (2019, background): “Team rewards can be difficult to co-optimize due to complex credit assignment, and can result in degenerate behavior where one agent learns a reasonable policy before its teammate, discouraging exploration which could interfere with the first agent’s behavior as observed by Hausknecht (2016).”
Value-Decomposition Networks For Cooperative Multi-Agent Learning Based On Team Reward (2018, background): “Further, in the context of the introduced agent, we evaluate weight sharing, role information and communication channels as additional enhancements that have recently been reported to improve sample complexity and memory requirements [8, 11, 22].”
Randomized Entity-wise Factorization for Multi-Agent Reinforcement Learning (2020, methodology): “Our approach utilizes value function factorization, an approach aiming to simultaneously overcome limitations of both joint (Hausknecht, 2016) and independent learning (Claus & Boutilier, 1998) paradigms.”
Arena: A General Evaluation Platform and Building Toolkit for Multi-Agent Intelligence (2019, unknown): “Other ideas include using joint action-value function [17, 26, 49, 54] and value function factorization [27, 44, 77], addressing the variance problem by a large batch size [5], learning grounded cooperative communication protocols between agents [18, 22, 30, 40, 50, 62, 89].”