Variational Dynamic for Self-Supervised Exploration in Deep Reinforcement Learning

Chenjia Bai

doi:10.1109/TNNLS.2021.3129160

Variational Dynamic for Self-Supervised Exploration in Deep Reinforcement Learning

2023-08-01

10

OA

AI

DOI：10.1109/TNNLS.2021.3129160

Original

Original request for help

Save

Abstract

En 中文

Efficient exploration remains a challenging problem in reinforcement learning, especially for tasks where extrinsic rewards from environments are sparse or even totally disregarded. Significant advances based on intrinsic motivation show promising results in simple environments but often get stuck in environments with multimodal and stochastic dynamics. In this work, we propose a variational dynamic model based on the conditional variational inference to model the multimodality and stochasticity. We consider the environmental state-action transition as a conditional generative process by generating the next-state prediction under the condition of the current state, action, and latent variable, which provides a better understanding of the dynamics and leads to a better performance in exploration. We derive an upper bound of the negative log likelihood of the environmental transition and use such an upper bound as the intrinsic reward for exploration, which allows the agent to learn skills by self-supervised exploration without observing extrinsic rewards. We evaluate the proposed method on several image-based simulation tasks and a real robotic manipulating task. Our method outperforms several state-of-the-art environment model-based exploration approaches.

Keywords:

Task analysis

Games

Estimation

Upper bound

Robots

Reinforcement learning

Noise measurement

Intrinsic motivation

reinforcement learning (RL)

self-supervised exploration

variational dynamic model (VDM)

Variational Dynamic for Self-Supervised Exploration in Deep Reinforcement Learning

Abstract

Journal

Organization