ModelBased Active Exploration
[edit]
Proceedings of the 36th International Conference on Machine Learning, PMLR 97:57795788, 2019.
Abstract
Efficient exploration is an unsolved problem in Reinforcement Learning which is usually addressed by reactively rewarding the agent for fortuitously encountering novel situations. This paper introduces an efficient active exploration algorithm, ModelBased Active eXploration (MAX), which uses an ensemble of forward models to plan to observe novel events. This is carried out by optimizing agent behaviour with respect to a measure of novelty derived from the Bayesian perspective of exploration, which is estimated using the disagreement between the futures predicted by the ensemble members. We show empirically that in semirandom discrete environments where directed exploration is critical to make progress, MAX is at least an order of magnitude more efficient than strong baselines. MAX scales to highdimensional continuous environments where it builds taskagnostic models that can be used for any downstream task.
Related Material


