Using previous models to bias structural learning in the hierarchical BOA

Hauschild, Mark W.; Pelikán, Martin; Sastry, Kumara; Goldberg, David E.

doi:10.1145/1389095.1389172

Cited by 35 publications

(31 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…That is why much effort has been put into enhancing efficiency of model building in EDAs and improving quality of EDA models even with smaller populations [6,8,9,20,21]. Learning from experience [5,6,12,20,21] represents one approach to addressing this issue.…”

Section: Learning From Experience Using Distance-based Biasmentioning

confidence: 99%

“…The definition of a distance between two variables of an ADF used in this paper as well as ref. [12] follows the work of Hauschild et al [6,11,20]. Given an ADF, we define the distance between two variables using a graph G of n nodes, one node per variable.…”

Section: Distance Metric For Additively Decomposable Functionsmentioning

confidence: 99%

“…This section describes the approach to learning from experience developed by Pelikan and Hauschild [12] inspired mainly by the work of Hauschild et al [6,20,21]. Let us assume a set M of hBOA models from prior hBOA runs on similar problems.…”

Section: Distance-based Bias Based On Previous Runs Of Hboamentioning

confidence: 99%

“…The use of probabilistic models in EDAs provides a basis for incorporating prior knowledge about the problem and learning from previous runs in order to solve new problem instances of similar type with increased speed, accuracy and reliability [5,6]. However, much prior work in this area was based on hand-crafted constraints on probabilistic models [7,8,9,10] which may be difficult to design or even detrimental to EDA efficiency and scalability [11].…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Transfer Learning, Soft Distance-Based Bias, and the Hierarchical BOA

Pelikán

Hauschild

Lanzi

2012

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

An automated technique has recently been proposed to transfer learning in the hierarchical Bayesian optimization algorithm (hBOA) based on distance-based statistics. The technique enables practitioners to improve hBOA efficiency by collecting statistics from probabilistic models obtained in previous hBOA runs and using the obtained statistics to bias future hBOA runs on similar problems. The purpose of this paper is threefold: (1) test the technique on several classes of NP-complete problems, including MAXSAT, spin glasses and minimum vertex cover; (2) demonstrate that the technique is effective even when previous runs were done on problems of different size; (3) provide empirical evidence that combining transfer learning with other efficiency enhancement techniques can often yield nearly multiplicative speedups. KeywordsTransfer learning, inductive transfer, learning from experience, estimation of distribution algorithms, hierarchical Bayesian optimization algorithm, decomposable problems, efficiency enhancement. Abstract An automated technique has recently been proposed to transfer learning in the hierarchical Bayesian optimization algorithm (hBOA) based on distance-based statistics. The technique enables practitioners to improve hBOA efficiency by collecting statistics from probabilistic models obtained in previous hBOA runs and using the obtained statistics to bias future hBOA runs on similar problems. The purpose of this paper is threefold: (1) test the technique on several classes of NP-complete problems, including MAXSAT, spin glasses and minimum vertex cover; (2) demonstrate that the technique is effective even when previous runs were done on problems of different size; (3) provide empirical evidence that combining transfer learning with other efficiency enhancement techniques can often yield nearly multiplicative speedups.

show abstract

Section: Learning From Experience Using Distance-based Biasmentioning

confidence: 99%

Section: Distance Metric For Additively Decomposable Functionsmentioning

confidence: 99%

Section: Distance-based Bias Based On Previous Runs Of Hboamentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Transfer Learning, Soft Distance-Based Bias, and the Hierarchical BOA

Pelikán

Hauschild

Lanzi

2012

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

show abstract

“…More recently, some work analyzed how different EDA components influence dependencies Lima et al 2007) and the use of the probabilistic models output by EDAs to speed up the solution of similar problems (Hauschild et al 2008). …”

Section: Analysis and Visualization Of Dependenciesmentioning

confidence: 99%

Mateda-2.0: AMATLABPackage for the Implementation and Analysis of Estimation of Distribution Algorithms

Santana¹,

Bielza²,

Larrañaga³

et al. 2010

J. Stat. Soft.

View full text Add to dashboard Cite

This paper describes Mateda-2.0, a MATLAB package for estimation of distribution algorithms (EDAs). This package can be used to solve single and multi-objective discrete and continuous optimization problems using EDAs based on undirected and directed probabilistic graphical models. The implementation contains several methods commonly employed by EDAs. It is also conceived as an open package to allow users to incorporate different combinations of selection, learning, sampling, and local search procedures. Additionally, it includes methods to extract, process and visualize the structures learned by the probabilistic models. This way, it can unveil previously unknown information about the optimization problem domain. Mateda-2.0 also incorporates a module for creating and validating function models based on the probabilistic models learned by EDAs.

show abstract