東華大學圖書館 |

Language: English

Help

回圖書館首頁

手機版館藏查詢

Back

Switch To: Labeled | MARC Mode | ISBD

Efficient reasoning in graphical models.

Rish, Irina.

Linked to FindBook

Google Book

Amazon

博客來

Efficient reasoning in graphical models.

Record Type:	Electronic resources : Monograph/item
Title/Author:	Efficient reasoning in graphical models./
Author:	Rish, Irina.
Description:	223 p.
Notes:	Source: Dissertation Abstracts International, Volume: 60-06, Section: B, page: 2790.
Contained By:	Dissertation Abstracts International60-06B.
Subject:	Computer Science. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=9934828
ISBN:	0599357215

Efficient reasoning in graphical models.
Rish, Irina.

Efficient reasoning in graphical models. - 223 p.

Source: Dissertation Abstracts International, Volume: 60-06, Section: B, page: 2790.

Thesis (Ph.D.)--University of California, Irvine, 1999.

Most artificial intelligence problems are computationally hard (NP-hard). However, in practice, the pessimistic worst-case performance can be improved by exploiting problem structure and by using approximations that trade accuracy for efficiency. Theoretical studies identify tractable problem classes while empirical evaluations shed light on average performance.

ISBN: 0599357215Subjects--Topical Terms:

626642
Computer Science.

Efficient reasoning in graphical models.
LDR:02378nmm 2200289 4500 001 1865127
005 20041216142214.5
008 130614s1999 eng d
020 $a 0599357215
035 $a (UnM)AAI9934828
035 $a AAI9934828
040 $a UnM $c UnM
100 1 $a Rish, Irina. $3 1952588
245 1 0 $a Efficient reasoning in graphical models.
300 $a 223 p.
500 $a Source: Dissertation Abstracts International, Volume: 60-06, Section: B, page: 2790.
500 $a Chair: Rina Dechter.
502 $a Thesis (Ph.D.)--University of California, Irvine, 1999.
520 $a Most artificial intelligence problems are computationally hard (NP-hard). However, in practice, the pessimistic worst-case performance can be improved by exploiting problem structure and by using approximations that trade accuracy for efficiency. Theoretical studies identify tractable problem classes while empirical evaluations shed light on average performance.
520 $a This thesis is concerned with efficient algorithms for automated inference in graphical models, such as constraint networks and belief networks. We use a general graph-based algorithmic framework that combines a dynamic-programming approach called variable elimination with conditioning techniques, such as backtracking search. We investigate the effects of certain problem structures, identify new tractable classes, and propose several structure-exploiting algorithms.
520 $a The central idea of this thesis is that efficiency can be gained by reducing the induced width, a graph parameter that bounds the complexity of variable elimination. We approach this problem by combining elimination with conditioning, which reduces the graph connectivity; by exploiting hidden structure such as causal independence in belief networks, which allows decomposition of large dependencies into smaller ones; and by using approximation algorithms that bound the size of recorded dependencies. Our empirical studies demonstrate promising results obtained both on randomly generated problems and on realistic domains such as medical diagnosis and probabilistic decoding.
590 $a School code: 0030.
650 4 $a Computer Science. $3 626642
690 $a 0984
710 2 0 $a University of California, Irvine. $3 705821
773 0 $t Dissertation Abstracts International $g 60-06B.
790 1 0 $a Dechter, Rina, $e advisor
790 $a 0030
791 $a Ph.D.
792 $a 1999
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=9934828