Trust-PCL: An Off-Policy Trust Region Method for Continuous Control.

Ofir Nachum Mohammad Norouzi Kelvin Xu Dale Schuurmans

Published in: CoRR (2017)

Keyphrases

trust region
objective function
cost function
closed form
training process
hessian matrix
dynamic programming
least squares
newton method
maximum likelihood
optimization algorithm
information theoretic
sensitivity analysis
global optimum