Learning algorithms for finite horizon constrained Markov decision processes

A. Mittal; N. Hemachandra

doi:10.3934/jimo.2007.3.429

Article Contents

2007, Volume 3, Issue 3: 429-444. Doi: 10.3934/jimo.2007.3.429

This issue Previous Article Conjugate duality for generalized convex optimization problems Next Article A solution framework for scheduling a BPM with non-identical job dimensions

Learning algorithms for finite horizon constrained Markov decision processes

A. Mittal^1, and
N. Hemachandra^1,

1.
IE and OR Interdisciplinary Programme, IIT Bombay, Mumbai, 400 076

Received: March 31, 2006

Revised: October 31, 2006

Published: June 2007

Abstract Related Papers Cited by

Abstract

We propose a heuristic and two stochastic approximation based learning algorithms for finite horizon, finite state-action constrained Markov decision models. We include models and numerical examples arising from risk management in fund allocation, retailer-depot product availability in a supply chain and admission control in a simple queue, that have to satisfy performance based constraints.

Keywords:

Mathematics Subject Classification: Primary: 90B50; Secondary: 11K45, 90C40.

Citation:

Article Metrics

HTML views() PDF downloads(153) Cited by(0)

Learning algorithms for finite horizon constrained Markov decision processes

Abstract

Access History

Article Metrics

Access History

Other Articles By Authors

Catalog

Learning algorithms for finite horizon constrained Markov decision processes

Abstract

Access History

SHARE

Article Metrics

Access History

Other Articles By Authors

Catalog

Export File

Citation

Format

Content