Actor-Critic or Critic-Actor? A Tale of Two Time Scales

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Bhatnagar, Shalabh, Borkar, Vivek S., Guin, Soumyajit
Format: Preprint
Published: 2022
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911924583661568
author Bhatnagar, Shalabh
Borkar, Vivek S.
Guin, Soumyajit
author_facet Bhatnagar, Shalabh
Borkar, Vivek S.
Guin, Soumyajit
contents We revisit the standard formulation of tabular actor-critic algorithm as a two time-scale stochastic approximation with value function computed on a faster time-scale and policy computed on a slower time-scale. This emulates policy iteration. We observe that reversal of the time scales will in fact emulate value iteration and is a legitimate algorithm. We provide a proof of convergence and compare the two empirically with and without function approximation (with both linear and nonlinear function approximators) and observe that our proposed critic-actor algorithm performs on par with actor-critic in terms of both accuracy and computational effort.
format Preprint
id arxiv_https___arxiv_org_abs_2210_04470
institution arXiv
publishDate 2022
record_format arxiv
spellingShingle Actor-Critic or Critic-Actor? A Tale of Two Time Scales
Bhatnagar, Shalabh
Borkar, Vivek S.
Guin, Soumyajit
Machine Learning
We revisit the standard formulation of tabular actor-critic algorithm as a two time-scale stochastic approximation with value function computed on a faster time-scale and policy computed on a slower time-scale. This emulates policy iteration. We observe that reversal of the time scales will in fact emulate value iteration and is a legitimate algorithm. We provide a proof of convergence and compare the two empirically with and without function approximation (with both linear and nonlinear function approximators) and observe that our proposed critic-actor algorithm performs on par with actor-critic in terms of both accuracy and computational effort.
title Actor-Critic or Critic-Actor? A Tale of Two Time Scales
topic Machine Learning
url https://arxiv.org/abs/2210.04470