Timezone: »

 
Efficient Planning in Large MDPs with Weak Linear Function Approximation - Csaba Szepesvari
Csaba Szepesvari

Fri Jul 17 04:00 PM -- 04:45 PM (PDT) @
in Theoretical Foundations of Reinforcement Learning »

Large-scale Markov decision processes (MDPs) require planning algorithms with runtime independent of the number of states of the MDP. We consider the planning problem in MDPs using linear value function approximation with only weak requirements: low approximation error for the optimal value function, and a small set of "core" states whose features span those of other states. In particular, we make no assumptions about the representability of policies or value functions of non-optimal policies. Our algorithm produces almost-optimal actions for any state using a generative oracle (simulator) for the MDP, while its computation time scales polynomially with the number of features, core states, and actions and the effective horizon. I will discuss how this is achieved, some selected part of the vast related literature and what remains open.

Joint work with Roshan Shariff

Author Information

Csaba Szepesvari (Deepmind)

More from the Same Authors