An Improved LAZY-AR Approach to Bayesian Network Inference

Butz, Cory J.; Song, Hua

doi:10.1007/11766247_16

Cited by 7 publications

(5 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The messages we identify are precisely those that will be propagated, provided that AR is chosen as the algorithm for building messages. However, (Butz and Hua 2006) pointed out that during its execution distributions can be built that will not be passed as messages, nor will they be needed in the construction of the distributions to be passed as messages. The reason is that in the elimination of a variable v i , AR constructs 3k − 1 distributions and outputs k of them, where k is the number of children of v i with respect to the distributions at the sending join tree node.…”

Section: Advantagesmentioning

confidence: 99%

Join tree propagation utilizing both arc reversal and variable elimination

Butz

Konkel

Lingras

2011

International Journal of Approximate Reasoning

View full text Add to dashboard Cite

In this paper, we put forth the first join tree propagation algorithm that selectively applies either arc reversal (AR) or variable elimination (VE) to build the propagated messages. Our approach utilizes a recent method for identifying the propagated join tree messagesà priori. When it is determined that precisely one message is to be constructed at a join tree node, VE is utilized to build this distribution; otherwise, AR is applied as it is better suited to construct multiple distributions passed between neighboring join tree nodes. Experimental results, involving evidence processing in seven real-world and one benchmark Bayesian network, empirically demonstrate that selectively applying VE and AR is faster than applying one of these methods exclusively on the entire network.

show abstract

Section: Advantagesmentioning

confidence: 99%

Join tree propagation utilizing both arc reversal and variable elimination

Butz

Konkel

Lingras

2011

International Journal of Approximate Reasoning

View full text Add to dashboard Cite

show abstract

“…Our approach uses VE to build distributions in computer memory. AR is not well suited for message construction in prioritized join tree propagation, because it constructs a new probability distribution for every child of the variable being eliminated [3]. As some of these new distributions can be deemed relevant and the rest irrelevant, we instead call VE to build the relevant distributions as needed.…”

Section: Physical Construction Of Messagesmentioning

confidence: 99%

“…It is also worth contrasting our work here with our previous work. In [3], we suggest applying the VE algorithm to build the messages in a Lazy AR schedule rather than applying the AR algorithm to construct the messages as Lazy AR does. We explicitly demonstrated that the AR algorithm can build probability distributions that will not be passed as messages, nor are they needed in the construction of the messages that will be passed.…”

Section: Examplementioning

confidence: 99%

Join tree propagation with prioritized messages

et al. 2009

Self Cite

View full text Add to dashboard Cite

Current join tree propagation algorithms treat all propagated messages as being of equal importance. On the contrary, it is often the case in real-world Bayesian networks that only some of the messages propagated from one join tree node to another are relevant to subsequent message construction at the receiving node. In this article, we propose the first join tree propagation algorithm that identifies and constructs the relevant messages first. Our approach assigns lower priority to the irrelevant messages as they only need to be constructed so that posterior probabilities can be computed when propagation terminates. Experimental results, involving the processing of evidence in four real-world Bayesian networks, empirically demonstrate an improvement over the stateof-the-art method for exact inference in discrete Bayesian networks.

show abstract

“…Given an input query p(X|Y = y), our system computes the posterior probabilities of the variables in X using our own inference algorithm, which is based on the recent work in [3]. More specifically, our algorithm is a hybrid approach utilizing join tree propagation to guide the inference procedure.…”

Section: A Prototype Interfacementioning

confidence: 99%

“…Of the three hybrid approaches tested, Lazy arc-reversal (Lazy-AR) was empirically shown to be the state-of-the-art method for exact inference in discrete BNs [7]. In [3], we proposed Lazy arc-reversal with variable elimination (Lazy-ARVE) as a new approach to BN inference and illustrated its benefits over Lazy-AR.…”

Section: Introductionmentioning

confidence: 99%

A Web-Based Interface for Hiding Bayesian Network Inference

Butz

Lingras

Konkel

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

Abstract. Bayesian networks have been applied for several uncertainty management problems in the artificial intelligence and Web intelligence communities. However, one may require the use of Bayesian networks, yet lack the background knowledge to build them. Moreover, it is widely acknowledged in the Bayesian network community that understanding Bayesian network inference is an arduous task. In this paper, we solve this dilemma by proposing a Web-based interface for hiding Bayesian network inference. This approach allows a much wider audience to utilize Bayesian network inference without having to understand how the inference process is actually carried out.

show abstract

An Improved LAZY-AR Approach to Bayesian Network Inference

Cited by 7 publications

References 14 publications

Join tree propagation utilizing both arc reversal and variable elimination

Join tree propagation utilizing both arc reversal and variable elimination

Join tree propagation with prioritized messages

A Web-Based Interface for Hiding Bayesian Network Inference

Contact Info

Product

Resources

About