Michael Xu

@mxu

I’m a researcher interested in simulation and AI. Currently I am trying to use reinforcement learning for control of deformable and amorphous objects.

Canada

Rocscience Inc.

University of Toronto

Michael's Spectra articles
Michael Xu
Four Identities for RL used in TRPO
Four Identities for RL used in TRPO
Presentation and proof of four identities related to advantage, value, and state-action value functions.
0 points
0 issues