Parallel Learning of Belief Networks in Large and Difficult Domains

Authors:
Y. Xiang;T. Chu
Affiliations:
Department of Computer Science, University of Regina, Regina, Saskatchewan, Canada S4S 0A2. yxiang@cs.uregina.ca;Avant Corporation, Sunnyvale, CA, USA
Venue:
Data Mining and Knowledge Discovery
Year:
1999

Citing 9
Cited 5

Probabilistic reasoning in intelligent systems: networks of plausible inference

Probabilistic reasoning in intelligent systems: networks of plausible inference
Introduction to parallel computing

Introduction to parallel computing
A Bayesian Method for the Induction of Probabilistic Networks from Data

Machine Learning
Learning Bayesian Networks: The Combination of Knowledge and Statistical Data

Machine Learning
A “Microscopic” Study of Minimum Entropy Search in Learning Decomposable Markov Networks

Machine Learning
Parallel Processing: From Applications to Systems

Parallel Processing: From Applications to Systems
Introduction to Bayesian Networks

Introduction to Bayesian Networks
an entropy-driven system for construction of probabilistic expert systems from databases

UAI '90 Proceedings of the Sixth Annual Conference on Uncertainty in Artificial Intelligence
Critical remarks on single link search in learning belief networks

UAI'96 Proceedings of the Twelfth international conference on Uncertainty in artificial intelligence

Data mining of Bayesian networks using cooperative coevolution

Decision Support Systems
Parallel BMDA with an aggregation of probability models

CEC'09 Proceedings of the Eleventh conference on Congress on Evolutionary Computation
Parallel learning of bayesian networks based on ordering of sets

ASIAN'05 Proceedings of the 10th Asian Computing Science conference on Advances in computer science: data management on the web
Review: learning bayesian networks: Approaches and issues

The Knowledge Engineering Review
Automatic construction of bayesian networks for conversational agent

ICIC'05 Proceedings of the 2005 international conference on Advances in Intelligent Computing - Volume Part II

Quantified Score

Hi-index	0.01

Visualization

Abstract

Learning belief networks from large domains can beexpensive even with single-link lookahead search (SLLS). Since aSLLS cannot learn correctly in a class of problem domains, multi-linklookahead search (MLLS) is needed which further increases thecomputational complexity. In our experiment, learning in somedifficult domains over more than a dozen variables took days. In thispaper, we study how to use parallelism to speed up SLLS for learningin large domains and to tackle the increased complexity of MLLS forlearning in difficult domains.We propose a natural decomposition of the learning task for parallelprocessing. We investigate two strategies for job allocation amongprocessors to further improve load balancing and efficiency of theparallel system. For learning from very large datasets, we present aregrouping of the available processors such that slow data accessthrough the file system can be replaced by fast memory access.Experimental results in a distributed memory MIMD computerdemonstrate the effectiveness of the proposed algorithms.