Time filter

Source Type

Aalborg, Denmark

Cabanas R.,University of Granada | Cano A.,University of Granada | Gomez-Olmedo M.,University of Granada | Madsen A.L.,HUGIN EXPERT A S | Madsen A.L.,University of Aalborg
International Journal of Approximate Reasoning | Year: 2016

An Influence Diagram is a probabilistic graphical model used to represent and solve decision problems under uncertainty. Its evaluation requires performing several combinations and marginalizations on the potentials attached to the Influence Diagram. Finding an optimal order for these operations, which is NP-hard, is an element of crucial importance for the efficiency of the evaluation. In this paper, two methods for optimizing this order are proposed. The first one is an improvement of the Variable Elimination algorithm while the second is the adaptation of the Symbolic Probabilistic Inference for evaluating Influence Diagrams. Both algorithms can be used for the direct evaluation of IDs but also for the computation of clique-to-clique messages in Lazy Evaluation of Influence Diagrams. In the experimental work, the efficiency of these algorithms is tested with several Influence Diagrams from the literature. © 2015 Elsevier Inc. All rights reserved.

Kaya B.,Millennium Villages Project | Madsen A.L.,HUGIN EXPERT A S
Agroforestry Systems | Year: 2016

This article discusses the potential of BNs to complement the analytical toolkit of agricultural extension. Statistical modelling of the adoption of agricultural practices has tended to use categorical (logit/probit) regression models focusing on a single technology or practice, explained by a number of household and farm characteristics. Here, a Bayesian network (BN) is used to model household-level data on adoption of agrosilvopastoral practices in Tiby, Mali. We discuss the advantages of BNs in modelling more complex data structures, including (i) multiple practices implemented jointly on farms, (ii) correlation between probabilities of implementation of those practices and (iii) correlation between household and farm characteristics. This paper demonstrates the use of BNs for ‘deductive’ reasoning regarding adoption of practices, answering questions regarding the probability of implementation of combinations of practices, conditional on household characteristics. As such, BNs is a complementary modelling approach to logistic regression analysis, which facilitates exploring causal structures in the data before deciding on a reduced form regression model. More uniquely, BNs can be used ‘inductively’ to answer questions regarding the likelihood of certain household characteristics conditional on certain practices being adopted. © 2016 Springer Science+Business Media Dordrecht

International Journal of Approximate Reasoning | Year: 2010

Even though existing algorithms for belief update in Bayesian networks (BNs) have exponential time and space complexity, belief update in many real-world BNs is feasible. However, in some cases the efficiency of belief update may be insufficient. In such cases minor improvements in efficiency may be important or even necessary to make a task tractable. This paper introduces two improvements to the message computation in Lazy propagation (LP): (1) we introduce myopic methods for sorting the operations involved in a variable elimination using arc-reversal and (2) extend LP with the any-space property. The performance impacts of the methods are assessed empirically. © 2010 Elsevier Inc. All rights reserved.

Madsen A.L.,University of Aalborg | Sondberg-Jeppesen N.,HUGIN EXPERT A S | Lohse N.,Loughborough University | Sayed M.S.,Loughborough University
Proceeding - 2015 IEEE International Conference on Industrial Informatics, INDIN 2015 | Year: 2015

This paper describes an innovative modular component-based modelling approach for diagnostics and condition-monitoring of manufacturing equipment. The approach is based on the use of object-oriented Bayesian networks, which supports a natural decomposition of a large and complex system into a set of less complex components. The methodology consists of six steps supporting the development process: Begin, Design, Implement, Test, Analyse, and Deploy. The process is iterative and the steps should be repeated until a satisfactory model has been achieved. The paper describes the details of the methodology as well as illustrates the use of the component-based modelling approach on a linear axis used in manufacturing. This application demonstrates the power and flexibility of the approach for diagnostics and condition-monitoring and shows a significant potential of the approach for modular component-based modelling in manufacturing and other domains. © 2015 IEEE.

Madsen A.L.,HUGIN EXPERT A S | Madsen A.L.,University of Aalborg | Butz C.J.,University of Regina | Oliveira J.S.,University of Regina | Dos Santos A.E.,University of Regina
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2016

Simple Propagation (SP) is a new junction tree-based algorithm for probabilistic inference in discrete Bayesian networks. It is similar to Lazy Propagation, but uses a simpler approach to exploit the factorization during message computation. The message construction is based on a one-in, one-out-principle meaning a potential has at least one non-evidence variable in the separator and at least one non-evidence variable not in the separator. This paper considers the use of different tree structures to guide the message passing in SP and reports on an experimental analysis using a set of real-world Bayesian networks. © Springer International Publishing Switzerland 2016.

Discover hidden collaborations