EVaDE : Event-Based Variational Thompson Sampling for Model-Based Reinforcement Learning

TL;DR

This paper introduces EVaDE, a novel event-based variational approach that enhances model-based reinforcement learning by improving exploration through domain-informed convolutional layers, demonstrated on Atari games.

Contribution

The paper proposes EVaDE, a new variational distribution leveraging event-based convolutional layers for better exploration in object-based domains within MBRL.

Findings

EVaDE improves exploration efficiency in Atari games.

EVaDE-SimPLe outperforms baseline methods.

Event-based layers facilitate better variational Thompson sampling.

Abstract

Posterior Sampling for Reinforcement Learning (PSRL) is a well-known algorithm that augments model-based reinforcement learning (MBRL) algorithms with Thompson sampling. PSRL maintains posterior distributions of the environment transition dynamics and the reward function, which are intractable for tasks with high-dimensional state and action spaces. Recent works show that dropout, used in conjunction with neural networks, induces variational distributions that can approximate these posteriors. In this paper, we propose Event-based Variational Distributions for Exploration (EVaDE), which are variational distributions that are useful for MBRL, especially when the underlying domain is object-based. We leverage the general domain knowledge of object-based domains to design three types of event-based convolutional layers to direct exploration. These layers rely on Gaussian dropouts and are inserted between the layers of the deep neural network model to help facilitate variational Thompson sampling. We empirically show the effectiveness of EVaDE-equipped Simulated Policy Learning (EVaDE-SimPLe) on the 100K Atari game suite.