Least Absolute Policy Iteration--A Robust Approach to Value Function Approximation

Masashi SUGIYAMA  Hirotaka HACHIYA  Hisashi KASHIMA  Tetsuro MORIMURA  

IEICE TRANSACTIONS on Information and Systems   Vol.E93-D    No.9    pp.2555-2565
Publication Date: 2010/09/01
Online ISSN: 1745-1361
DOI: 10.1587/transinf.E93.D.2555
Print ISSN: 0916-8532
Type of Manuscript: PAPER
Category: Artificial Intelligence, Data Mining
reinforcement learning,  value function approximation,  least-squares policy iteration,  outlier,  1-loss function,  linear programming,  

Full Text: PDF>>
Buy this Article

Least-squares policy iteration is a useful reinforcement learning method in robotics due to its computational efficiency. However, it tends to be sensitive to outliers in observed rewards. In this paper, we propose an alternative method that employs the absolute loss for enhancing robustness and reliability. The proposed method is formulated as a linear programming problem which can be solved efficiently by standard optimization software, so the computational advantage is not sacrificed for gaining robustness and reliability. We demonstrate the usefulness of the proposed approach through a simulated robot-control task.