Record number :
1675604
Title of article :
An actor–critic algorithm with function approximation for discounted cost constrained Markov decision processes
Author/Authors :
Bhatnagar، نويسنده , , Shalabh، نويسنده ,
Issue Information :
ماهنامه با شماره پیاپی سال 2010
Pages :
7
From page :
760
To page :
766
Abstract :
We develop in this article the first actor–critic reinforcement learning algorithm with function approximation for a problem of control under multiple inequality constraints. We consider the infinite horizon discounted cost framework in which both the objective and the constraint functions are suitable expected policy-dependent discounted sums of certain sample path functions. We apply the Lagrange multiplier method to handle the inequality constraints. Our algorithm makes use of multi-timescale stochastic approximation and incorporates a temporal difference (TD) critic and an actor that makes a gradient search in the space of policy parameters using efficient simultaneous perturbation stochastic approximation (SPSA) gradient estimates. We prove the asymptotic almost sure convergence of our algorithm to a locally optimal policy.
Keywords :
Infinite horizon discounted cost criterion , function approximation , Simultaneous perturbation stochastic approximation , Actor–critic algorithm , Constrained Markov decision processes
Journal title :
Systems and Control Letters
Serial Year :
2010
Link To Document :
بازگشت