|Speaker :||Sean Meyn|
|University of Florida|
|Time:||2:00 pm - 3:00 pm|
|Location:||LINCS Meeting Room 40|
Neuro-Dynamic ProgrammingÂ encompasses techniques from both reinforcement learning and approximateÂ dynamic programming. Feature selection refers to the choice of basis that defines the function class that isÂ required in the application of these techniques. This talk reviews two popular approaches to neuro-dynamicÂ programming, TD-learning and Q-learning. The main goal of this work is to demonstrate how insight fromÂ idealized models can be used as a guide for feature selection for these algorithms. Several approaches areÂ surveyed, including fluid and diffusion models, and the application of idealized models arising from mean-fieldÂ game approximations. The theory is illustrated with several examples.