I presented a livestream, available here, which is meant as part 2 of a series on Bayesian Reinforcement Learning, but this one focuses on Gaussian processes (as a prelude to discussing Gaussian Process Temporal Difference learning in an upcoming presentation).
Of course, the code examples I use are in PyMC3.
- Quick recap of part 1 (Bayesian Q-Learning and VPI)
- Intro to Gaussian processes
- Intro to kernel functions
- Code example
- A few thoughts on good hypotheses and data inference
I hope this will be of interest to someone here.