End-to-End Policy Gradient Method for POMDPs and Explainable Agents

Soichiro Nishimori, Sotetsu Koyamada, Shin Ishii

Real-world decision-making problems are often partially observable, and many can be formulated as a Partially Observable Markov Decision Process (POMDP). When we apply reinforcemen…