This is one of the important paper that link MBRL with Variational Inference, published in 2018/05
Problem:
the connection between reinforcement learning and inference in probabilistic models is not immediately obvious
Innovation:
Discuss how a generalization of the reinforcement learning or optimal control problem is equivalent to exact probabilistic inference in the case of deterministic dynamics, and variational inference in the case of stochastic dynamics
Conclusion/Application:
intersection of maximum entropy reinforcement learning and latent variable models
design of reward functions
Comment:
This paper put some theory background to connect RL with Statistical Inference. It first bring up the PGM (Probability Graph Model). Then in "Policy Search as Probabilistic Inference" section, Q/V value is defined by analogously to inference in HMM-style dynamic Bayesian networks
Link: Arxiv
This is one of the important paper that link MBRL with Variational Inference, published in 2018/05
Problem:
Innovation:
Conclusion/Application:
Comment: