Linear complementarity for regularized policy evaluation and improvement
Recent work in reinforcement learning has emphasized the power of L 1 regularization to perform feature selection and prevent overfitting. We propose formulating the L1 regularized linear fixed point problemas a linear complementarity problem (LCP). This formulation offers several advantages over the LARS-inspired formulation, LARS-TD. The LCP formulation allows the use of efficient off-theshelf solvers, leads to a new uniqueness result, and can be initialized with starting points from similar problems (warm starts). We demonstrate that warm starts, as well as the efficiency of LCP solvers, can speed up policy iteration. Moreover, warm starts permit a form of modified policy iteration that can be used to approximate a "greedy" homotopy path, a generalization of the LARS-TD homotopy path that combines policy evaluation and optimization.
