Maximum Entropy Semi-Supervised Inverse Reinforcement Learning

Abstract

A popular approach to apprenticeship learning (AL) is to formulate it as an inverse reinforce-ment learning (IRL) problem. The MaxEnt-IRL al-gorithm successfully integrates the maximum en-tropy principle into IRL and unlike its predecessors, it resolves the ambiguity arising from the fact that a possibly large number of policies could match the expert’s behavior. In this paper, we study an AL setting in which in addition to the expert’s tra-jectories, a number of unsupervised trajectories is available. We introduce MESSI, a novel algorithm that combines MaxEnt-IRL with principles com-ing from semi-supervised learning. In particular, MESSI integrates the unsupervised data into the MaxEnt-IRL framework using a pairwise penalty on trajectories. Empirical results in a highway driv-ing and grid-world problems indicate that MESSI is able to take advantage of the unsupervised trajecto-ries and improve the performance of MaxEnt-IRL. 1