Of Uncertain Value -- Bayesian Quadratic Reinforcement Learning
Add to your list(s)
Download to your calendar using vCal
If you have a question about this talk, please contact Emli-Mari Nel.
How much do you need to know about the future to take the right decisions?
In this talk, I will briefly review two influential recent works in Bayesian Reinforcement Learning, and then move on to my own work in progress, on an attempt to find a good Gaussian approximation to the Bellman posterior over values. Unfortunately, the “right” Gaussian fit turns out to be very difficult to track, but a simplistic approximation gives surprisingly good results. Preliminary experiments suggest that this approach compares favorably to contemporary Reinforcement Learning algorithms.
This talk is part of the Inference Group series.
This talk is included in these lists:
Note that ex-directory lists are not shown.
|