Continuous Coordination As a Realistic Scenario for Lifelong Learning

Abstract

Current deep reinforcement learning (RL) algorithms are still highly task-specific and lack the ability to generalize to new environments. Lifelong learning (LLL), however, aims at solving multiple tasks sequentially by efficiently transferring and using knowledge between tasks. Despite a surge of interest in lifelong RL in recent years, the lack of a realistic testbed makes robust evaluation of LLL algorithms difficult. Multi-agent RL (MARL), on the other hand, can be seen as a natural scenario for lifelong RL due to its inherent non-stationarity, since the agents' policies change over time. In this work, we introduce a multi-agent lifelong learning testbed that supports both zero-shot and few-shot settings. Our setup is based on Hanabi -- a partially-observable, fully cooperative multi-agent game that has been shown to be challenging for zero-shot coordination. Its large strategy space makes it a desirable environment for lifelong RL tasks. We evaluate several recent MARL methods, and benchmark state-of-the-art LLL algorithms in limited memory and computation regimes to shed light on their strengths and weaknesses. This continual learning paradigm also provides us with a pragmatic way of going beyond centralized training which is the most commonly used training protocol in MARL. We empirically show that the agents trained in our setup are able to coordinate well with unseen agents, without any additional assumptions made by previous works. The code and all pre-trained models are available at https://github.com/chandar-lab/Lifelong-Hanabi.

arXiv comment19 pages with supplementary materials. Added results for Lifelong RL methods and some future work. Accepted to ICML 2021

References in corpus (25)

• Overcoming catastrophic forgetting in neural networks
• Playing Atari with Deep Reinforcement Learning
• Prioritized Experience Replay
• Theoretical Models of Learning to Learn
• Riemannian Walk for Incremental Learning: Understanding Forgetting and Intransigence
• Continual Learning Through Synaptic Intelligence
• An Empirical Investigation of Catastrophic Forgetting in Gradient-Based Neural Networks
• Efficient Lifelong Learning with A-GEM
• Learning a Generic Value-Selection Heuristic Inside a Constraint Programming Solver
• On Tiny Episodic Memories in Continual Learning
• Invariant Risk Minimization
• Progress & Compress: A scalable framework for continual learning
• The Hanabi Challenge: A New Frontier for AI Research
• Cooperating with Machines
• Deep Decentralized Multi-task Multi-Agent Reinforcement Learning under Partial Observability
• Fast Context Adaptation via Meta-Learning
• Continuous Adaptation via Meta-Learning in Nonstationary and Competitive Environments
• A Dissection of Overfitting and Generalization in Continuous Reinforcement Learning
• Bayesian Action Decoder for Deep Multi-Agent Reinforcement Learning
• Simplified Action Decoder for Deep Multi-Agent Reinforcement Learning
• A Deep Policy Inference Q-Network for Multi-Agent Systems
• "Other-Play" for Zero-Shot Coordination
• Defining Benchmarks for Continual Few-Shot Learning
• Benchmark Environments for Multitask Learning in Continuous Domains
• Jelly Bean World: A Testbed for Never-Ending Learning