On Neural Consolidation for Transfer in Reinforcement Learning

Date:

October 5, 2022

2022

Type:

Conference

Publication:

IEEE Symposium on Adaptive Dynamic Programming and Reinforcement Learning

Author(s):

Valentin Guillet, Dennis G. Wilson, Carlos Aguilar Melchor, Emmanuel Rachelson

Topic:

Abstract

Although transfer learning is considered to be a milestone in deep reinforcement learning, the mechanisms behind it are still poorly understood. In particular, predicting if knowledge can be transferred between two given tasks is still an unresolved problem.

In this work, we explore the use of network distillation as a feature extraction method to better understand the context in which transfer can occur. Notably, we show that distillation does not prevent knowledge transfer, including when transferring from multiple tasks to a new one, and we compare these results with transfer without prior distillation. We focus our work on the Atari benchmark due to the variability between different games, but also to their similarities in terms of visual features.

Download Paper
Back to all publications