Reinforcement Learning When All Actions are Not Always Available
This addresses a specific limitation in RL for real-world sequential decision-making where action availability varies stochastically, offering a more stable solution.
The paper tackles the problem of reinforcement learning when actions are not always available, known as stochastic action set MDPs, by proposing new policy gradient algorithms with variance reduction techniques and proving their convergence, demonstrating practicality on real-life inspired tasks.
The Markov decision process (MDP) formulation used to model many real-world sequential decision making problems does not efficiently capture the setting where the set of available decisions (actions) at each time step is stochastic. Recently, the stochastic action set Markov decision process (SAS-MDP) formulation has been proposed, which better captures the concept of a stochastic action set. In this paper we argue that existing RL algorithms for SAS-MDPs can suffer from potential divergence issues, and present new policy gradient algorithms for SAS-MDPs that incorporate variance reduction techniques unique to this setting, and provide conditions for their convergence. We conclude with experiments that demonstrate the practicality of our approaches on tasks inspired by real-life use cases wherein the action set is stochastic.