Belief network algorithms: A study of performance based on domain characterisation

Jitnah, Nathalie; Nicholson, Ann E.

doi:10.1007/3-540-64413-x_35

Cited by 11 publications

(8 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These particular features were chosen according to the domain knowledge, previous literature (Jitnah and Nicholson 1998;Ide and Cozman 2002;Shimony and Domshlak 2003), and our initial experimental results. Another practical reason is because they can all be calculated in polynomial time.…”

Section: The Algorithm Spacementioning

confidence: 99%

“…Since we only consider binary nodes, the maximum number of parents of a node, max_parents, can be used to bound the CPT size. The skewness of the CPTs is computed as follows (Jitnah and Nicholson 1998): for a vector (a column of the CPT table), v = (v 1 , v 2 , . .…”

Section: The Algorithm Spacementioning

confidence: 99%

“…In the future it is necessary to conduct a thorough experimental comparison between regression models and classifiers for algorithm selection. Other related works include (Gomes and Selman 1997;Jitnah and Nicholson 1998;Hoos and Stutzle 1998;Ruan et al 2004). …”

Section: Related Workmentioning

confidence: 99%

See 2 more Smart Citations

A machine learning approach to algorithm selection for $\mathcal{NP}$ -hard optimization problems: a case study on the MPE problem

Guo

Hsu

2007

Ann Oper Res

View full text Add to dashboard Cite

Given one instance of an N P-hard optimization problem, can we tell in advance whether it is exactly solvable or not? If it is not, can we predict which approximate algorithm is the best to solve it? Since the behavior of most approximate, randomized, and heuristic search algorithms for N P-hard problems is usually very difficult to characterize analytically, researchers have turned to experimental methods in order to answer these questions. In this paper we present a machine learning-based approach to address the above questions. Models induced from algorithmic performance data can represent the knowledge of how algorithmic performance depends on some easy-to-compute problem instance characteristics. Using these models, we can estimate approximately whether an input instance is exactly solvable or not. Furthermore, when it is classified as exactly unsolvable, we can select the best approximate algorithm for it among a list of candidates. In this paper we use the MPE (most probable explanation) problem in probabilistic inference as a case study to validate the proposed methodology. Our experimental results show that the machine learning-based algorithm selection system can integrate both exact and inexact algorithms and provide the best overall performance comparing to any single candidate algorithm.

show abstract

Section: The Algorithm Spacementioning

confidence: 99%

Section: The Algorithm Spacementioning

confidence: 99%

See 1 more Smart Citation

A machine learning approach to algorithm selection for $\mathcal{NP}$ -hard optimization problems: a case study on the MPE problem

Guo

Hsu

2007

Ann Oper Res

View full text Add to dashboard Cite

show abstract

“…In general, the complexity of both exact and approximate updating is NP-hard [8]. The exact inference algorithms [6] include the Kim-Pearl poly-tree algorithm and Spiegelhalter's clique tree algorithm.…”

Section: Dynamic Bayesian Network and Infer-ence Algorithmsmentioning

confidence: 99%

Applied probabilistic AI for online diagnosis of a safety-critical system based on a quality assurance program

Lauber

Steger

Weiß

1999

Proceedings of the 1999 ACM Symposium on Applied Computing

View full text Add to dashboard Cite

The goal of probabilistic Artificial Intelligence (Al) in diagnosis iS IO impIWe performance Offadl isolation in dynamic systems. However; this Al method is dependent on prior probabilisric fault knowledge. We present a hybrid technique in combining reliability engineering and probabilistic Al. We develop a Dynamic Bayesian Network (DBN) which, given evidence from sensor observations, is able to learn and reason over time. We successfully apply the DBN in concert with reliability assessmetu to improve online diagnosis of a pneumaticmechanical device.

show abstract

“…How to efficiently solve these problems is still an open question. The existing algorithms seldom or only superficially consider the structural characteristics of these problems 4 ' 5 ' 6 ' 7 , and this makes these algorithms sensitive to the solution landscape 8 ' 9 ' 10 . In this paper, we continue and extend the work on structure-based optimization 8 .…”

Section: Introductionmentioning

confidence: 99%

Genetic Algorithms and Reinforcement Learning for the Tactical Fixed Interval Scheduling Problem

Santos

Zhong

2001

Int. J. Artif. Intell. Tools

View full text Add to dashboard Cite

Discrete optimization problems are usually NP hard. The structural characteristics of these problems significantly dictate the solution landscape. In this paper, we explore a structure-based approach to solving these kinds of problems. We use a reinforcement learning system to adaptively learn the structural characteristics of the problem, hereby decomposing the problem into several subproblems. Based on these structural characteristics, we develop a Genetic Algorithm by using structural operations to recombine these subproblems together to solve the problem. The reinforcement learning system directs the GA. We test our algorithm on the Tactical Fixed Interval Scheduling Problem(TFISP) which is the problem of determining the minimum number of parallel non-identical machine such that a feasible schedule exists for a given set of jobs. This work continues our work in exploiting structure for optimization

show abstract

Belief network algorithms: A study of performance based on domain characterisation

Cited by 11 publications

References 10 publications

A machine learning approach to algorithm selection for $\mathcal{NP}$ -hard optimization problems: a case study on the MPE problem

A machine learning approach to algorithm selection for $\mathcal{NP}$ -hard optimization problems: a case study on the MPE problem

Applied probabilistic AI for online diagnosis of a safety-critical system based on a quality assurance program

Genetic Algorithms and Reinforcement Learning for the Tactical Fixed Interval Scheduling Problem

Contact Info

Product

Resources

About