Stochastic Shortest Path: Minimax, Parameter-Free and Towards Horizon-Free Regret
share
We study the problem of learning in the stochastic shortest path (SSP) setting, where an agent seeks to minimize the expected cost accumulated before reaching a goal state. We design a novel model-based algorithm EB-SSP that carefully skews the empirical transitions and perturbs the empirical costs with an exploration bonus to guarantee both optimism and convergence of the associated value iteration scheme. We prove that EB-SSP achieves the minimax regret rate O(B_⋆√(S A K)), where K is the number of episodes, S is the number of states, A is the number of actions and B_⋆ bounds the expected cumulative cost of the optimal policy from any state, thus closing the gap with the lower bound. Interestingly, EB-SSP obtains this result while being parameter-free, i.e., it does not require any prior knowledge of B_⋆, nor of T_⋆ which bounds the expected time-to-goal of the optimal policy from any state. Furthermore, we illustrate various cases (e.g., positive costs, or general costs when an order-accurate estimate of T_⋆ is available) where the regret only contains a logarithmic dependence on T_⋆, thus yielding the first horizon-free regret bound beyond the finite-horizon MDP setting.
READ FULL TEXT