A new book titled Simulation-based Algorithms for Markov Decision Processes was co-written by four authors with ties to the Electrical and Computer Engineering Department: former ECE Chair and current Professor Steve Marcus (ECE/ISR); Professor Michael Fu (BUS/ECE/ISR); former Institute for Systems Research (ISR) postdoctoral researcher Hyeong Soo Chang, who now teaches at Sogang University in Seoul, Korea; and ECE alumnus Jiaqiao Hu, who was advised by Marcus and Fu and is now an assistant professor at SUNY Stony Brook University.
The new publication, part of the Springer's Communications and Control Engineering series, explores Markov decision processes (MDP). MDPs are widely used for modeling sequential decision-making problems that arise in engineering, economics, computer science, and the social sciences. The book helps provide sampling and population-based numerical algorithms to overcome the difficulties of computing an optimal solution using MDP models in terms of a policy or value function.
Specific approaches used in the book include multi-stage adaptive sampling, evolutionary policy iteration, evolutionary random policy search, and model reference adaptive search. The book provides specific algorithms, illustrative numerical examples and rigorous theoretical convergence results. The new book should appeal not only to researchers in MDPs, stochastic modeling and control, and simulation but should also be a valuable reference for students of control and operations research.
For more information about the book, visit the Springer website.
February 26, 2007