Learning Approximate Stochastic Transition Models

Yuhang Song, Christopher Grimm, Xianming Wang +1

We examine the problem of learning mappings from state to state, suitable for use in a model-based reinforcement-learning setting, that simultaneously generalize to novel states an…