Stock Image

Learning Representation and Control in Markov Decision Processes: New Frontiers

Sridhar Mahadevan

0 ratings by Goodreads
ISBN 10: 1601982380 / ISBN 13: 9781601982384
Published by Now Publishers
New Condition: New Soft cover
Buy New
List Price: US$ 110.00
Price: US$ 182.86 Convert Currency
Shipping: US$ 3.95 Within U.S.A. Destination, Rates & Speeds
Add to basket

30 Day Return Policy

Offered By

BuySomeBooks
Las Vegas, NV, U.S.A.

5-star rating

AbeBooks Seller Since May 21, 2012

About this Item

Paperback. 184 pages. Dimensions: 9.2in. x 6.1in. x 0.4in.Learning Representation and Control in Markov Decision Processes describes methods for automatically compressing Markov decision processes (MDPs) by learning a low-dimensional linear approximation defined by an orthogonal set of basis functions. A unique feature of the text is the use of Laplacian operators, whose matrix representations have non-positive off-diagonal elements and zero row sums. The generalized inverses of Laplacian operators, in particular the Drazin inverse, are shown to be useful in the exact and approximate solution of MDPs. The author goes on to describe a broad framework for solving MDPs, generically referred to as representation policy iteration (RPI), where both the basis function representations for approximation of value functions as well as the optimal policy within their linear span are simultaneously learned. Basis functions are constructed by diagonalizing a Laplacian operator or by dilating the reward function or an initial set of bases by powers of the operator. The idea of decomposing an operator by finding its invariant subspaces is shown to be an important principle in constructing low-dimensional representations of MDPs. Theoretical properties of these approaches are discussed, and they are also compared experimentally on a variety of discrete and continuous MDPs. Finally, challenges for further research are briefly outlined. Learning Representation and Control in Markov Decision Processes is a timely exposition of a topic with broad interest within machine learning and beyond. This item ships from multiple locations. Your book may arrive from Roseburg,OR, La Vergne,TN. Bookseller Inventory # 9781601982384

Quantity Available: > 20

Ask Seller a Question

Bibliographic Details

Title: Learning Representation and Control in ...

Publisher: Now Publishers

Binding: Paperback

Book Condition: New

Book Type: Paperback

About this title

Synopsis:

Learning Representation and Control in Markov Decision Processes describes methods for automatically compressing Markov decision processes (MDPs) by learning a low-dimensional linear approximation defined by an orthogonal set of basis functions. A unique feature of the text is the use of Laplacian operators, whose matrix representations have non-positive off-diagonal elements and zero row sums. The generalized inverses of Laplacian operators, in particular the Drazin inverse, are shown to be useful in the exact and approximate solution of MDPs. The author goes on to describe a broad framework for solving MDPs, generically referred to as representation policy iteration (RPI), where both the basis function representations for approximation of value functions as well as the optimal policy within their linear span are simultaneously learned. Basis functions are constructed by diagonalizing a Laplacian operator or by dilating the reward function or an initial set of bases by powers of the operator. The idea of decomposing an operator by finding its invariant subspaces is shown to be an important principle in constructing low-dimensional representations of MDPs. Theoretical properties of these approaches are discussed, and they are also compared experimentally on a variety of discrete and continuous MDPs. Finally, challenges for further research are briefly outlined. Learning Representation and Control in Markov Decision Processes is a timely exposition of a topic with broad interest within machine learning and beyond.

About the Author:

University of Massachusetts, Amherst

"About this title" may belong to another edition of this title.

Store Description

BuySomeBooks is great place to get your books online. With over eight million titles available we're sure to have what you're looking for. Despite having a large selection of new books available for immediate shipment and excellent customer service, people still tell us they prefer us because of our prices.

Visit Seller's Storefront

Terms of Sale:

We guarantee the condition of every book as it's described on the Abebooks web
sites. If you're dissatisfied with your purchase (Incorrect Book/Not as
Described/Damaged) or if the order hasn't arrived, you're eligible for a refund
within 30 days of the estimated delivery date. If you've changed your mind about a book that you've ordered, please use the Ask bookseller a question link to contact us and we'll respond within 2 business days.

BuySomeBooks is operated by Drive-On-In, Inc., a Nevada co...

More Information
Shipping Terms:

Orders usually ship within 1-2 business days. Books are shipped from multiple locations so your order may arrive from Las Vegas,NV, Roseburg,OR, La Vergne,TN, Momence,IL, or Commerce,GA.

List this Seller's Books

Payment Methods
accepted by seller

Visa Mastercard American Express