Some Considerations on Learning to Explore via Meta-Reinforcement Learning

03/03/2018

∙

We consider the problem of exploration in meta reinforcement learning. Two new meta reinforcement learning algorithms are suggested: E-MAML and E-RL^2. Results are presented on a novel environment we call `Krazy World' and a set of maze environments. We show E-MAML and E-RL^2 deliver better performance on tasks where exploration is important.

READ FULL TEXT

Some Considerations on Learning to Explore via Meta-Reinforcement Learning

Meta-Reinforcement Learning for Heuristic Planning

A Brief Look at Generalization in Visual Meta-Reinforcement Learning

Improving Generalization in Meta Reinforcement Learning using Learned Objectives

REIN-2: Giving Birth to Prepared Reinforcement Learning Agents Using Reinforcement Learning Agents

A Survey of Exploration Methods in Reinforcement Learning

EVO-RL: Evolutionary-Driven Reinforcement Learning

A Meta-MDP Approach to Exploration for Lifelong Reinforcement Learning

Some Considerations on Learning to Explore via Meta-Reinforcement Learning

Related Research

Meta-Reinforcement Learning for Heuristic Planning

A Brief Look at Generalization in Visual Meta-Reinforcement Learning

Improving Generalization in Meta Reinforcement Learning using Learned Objectives

REIN-2: Giving Birth to Prepared Reinforcement Learning Agents Using Reinforcement Learning Agents

A Survey of Exploration Methods in Reinforcement Learning

EVO-RL: Evolutionary-Driven Reinforcement Learning

A Meta-MDP Approach to Exploration for Lifelong Reinforcement Learning