Poster

Provably Efficient Learning of Transferable Rewards

Alberto Maria Metelli · Giorgia Ramponi · Alessandro Concetti · Marcello Restelli

Virtual

Keywords: [ Combinatorial Optimization ] [ Convex Optimization ] [ Reinforcement Learning and Planning ]

[ Abstract ]
[ Slides
[ Paper ]
[ Visit Poster at Spot A0 in Virtual World ]
Tue 20 Jul 9 a.m. PDT — 11 a.m. PDT
 
Spotlight presentation: Reinforcement Learning 1
Tue 20 Jul 6 a.m. PDT — 7 a.m. PDT

Abstract:

The reward function is widely accepted as a succinct, robust, and transferable representation of a task. Typical approaches, at the basis of Inverse Reinforcement Learning (IRL), leverage on expert demonstrations to recover a reward function. In this paper, we study the theoretical properties of the class of reward functions that are compatible with the expert’s behavior. We analyze how the limited knowledge of the expert’s policy and of the environment affects the reward reconstruction phase. Then, we examine how the error propagates to the learned policy’s performance when transferring the reward function to a different environment. We employ these findings to devise a provably efficient active sampling approach, aware of the need for transferring the reward function, that can be paired with a large variety of IRL algorithms. Finally, we provide numerical simulations on benchmark environments.

Chat is not available.