Combining Reinformation Learning with Information-State Update Rules

Reinforcement learning gives a way to learn under what circumstances to perform which actions. However, this approach lacks a formal framework for specifying hand-crafted restrictions, for specifying the effects of the system actions, or for specifying the user simulation. The information state approach, in contrast, allows system and user behavior to be specified as update rules, with preconditions and effects. This approach can be used to specify complex dialogue behavior in a systematic way. We propose combining these two approaches, thus allowing a formal specification of the dialogue behavior, and allowing hand-crafted preconditions, with remaining ones determined via reinforcement learning so as to minimize dialogue cost.