MAME : Model-Agnostic Meta-Exploration

TL;DR

This paper introduces MAME, a model-agnostic meta-exploration framework that employs separate exploration and exploitation policies to improve adaptation efficiency in meta-reinforcement learning tasks.

Contribution

It proposes explicitly modeling separate exploration and exploitation policies, enhancing training flexibility and adaptation efficiency compared to prior methods.

Findings

Superior performance over prior methods in meta-reinforcement learning tasks

Effective use of self-supervised or supervised objectives for adaptation

More efficient inner-loop updates with separate exploration policy

Abstract

Meta-Reinforcement learning approaches aim to develop learning procedures that can adapt quickly to a distribution of tasks with the help of a few examples. Developing efficient exploration strategies capable of finding the most useful samples becomes critical in such settings. Existing approaches towards finding efficient exploration strategies add auxiliary objectives to promote exploration by the pre-update policy, however, this makes the adaptation using a few gradient steps difficult as the pre-update (exploration) and post-update (exploitation) policies are often quite different. Instead, we propose to explicitly model a separate exploration policy for the task distribution. Having two different policies gives more flexibility in training the exploration policy and also makes adaptation to any specific task easier. We show that using self-supervised or supervised learning objectives for adaptation allows for more efficient inner-loop updates and also demonstrate the superior performance of our model compared to prior works in this domain.