On Solving a Stochastic Shortest-Path Markov Decision Process as Probabilistic Inference
This addresses a generalization of MDPs for AI planning under uncertainty, but appears incremental as it builds on existing active inference approaches.
The paper tackles solving Stochastic Shortest-Path Markov Decision Processes (SSP MDPs) by framing them as probabilistic inference, extending prior work limited to finite horizons to handle indefinite horizons, and discusses online and offline planning methods.
Previous work on planning as active inference addresses finite horizon problems and solutions valid for online planning. We propose solving the general Stochastic Shortest-Path Markov Decision Process (SSP MDP) as probabilistic inference. Furthermore, we discuss online and offline methods for planning under uncertainty. In an SSP MDP, the horizon is indefinite and unknown a priori. SSP MDPs generalize finite and infinite horizon MDPs and are widely used in the artificial intelligence community. Additionally, we highlight some of the differences between solving an MDP using dynamic programming approaches widely used in the artificial intelligence community and approaches used in the active inference community.