Crossmodal attentive skill learner

Abstract

© 2018 International Foundation for Autonomous Agents and Multiagent Systems. This paper introduces the Crossmodal Attentive Skill Learner (CASL), Integrated with the recently-introduced Asynchronous Advantage Option-Critic (A2OC) architecture (15] to enable hierarchical rei nforcement learning across multiple sensory inputs. We provide concrete examples where the approach not only improves perform ance in a single task, but accelerates transfer to new tasks. We demonstrate the attention mechanism anticipates and identifies useful latent features, while filtering irrelevant sensor modalities during execution. We modify the Arcade Learning Environment (7] to support audio queries, and conduct evaluations of crossmodal learning in the Atari 2600 games H.E.R.O. and Amidar. Finally, buildi ng on the recent work of Babaeizadeh et aL [4], we open-soulce a fast hybrid CPU-CPU implementation of CASL.