Continuous Adaptation via Meta-Learning in Nonstationary and Competitive Environments

TL;DR

This paper introduces a gradient-based meta-learning approach for continuous adaptation in nonstationary, adversarial environments, validated through a new multi-agent competitive environment called RoboSumo.

Contribution

It presents a simple meta-learning algorithm tailored for dynamic, adversarial scenarios and introduces RoboSumo for testing continuous adaptation strategies.

Findings

Meta-learning enables more efficient adaptation than reactive baselines.

Meta-learners outperform in competitive multi-agent environments.

Agents that learn and adapt via meta-learning are more robust and fit.

Abstract

Ability to continuously learn and adapt from limited experience in nonstationary environments is an important milestone on the path towards general intelligence. In this paper, we cast the problem of continuous adaptation into the learning-to-learn framework. We develop a simple gradient-based meta-learning algorithm suitable for adaptation in dynamically changing and adversarial scenarios. Additionally, we design a new multi-agent competitive environment, RoboSumo, and define iterated adaptation games for testing various aspects of continuous adaptation strategies. We demonstrate that meta-learning enables significantly more efficient adaptation than reactive baselines in the few-shot regime. Our experiments with a population of agents that learn and compete suggest that meta-learners are the fittest.