Rule Extraction from Linear Support Vector Machines via Mathematical Programming

Fung, Glenn; Sandilya, Sathyakama; Rao, R. Bharat

doi:10.1007/978-3-540-75390-2_4

Cited by 7 publications

(5 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Now it has been proved to be a powerful and promising data classification and function estimation tool. Reference [2,3] applied SVM to data classification. They have obtained some valuable results.…”

Section: Introductionmentioning

confidence: 99%

Data Classification Using Support Vector Machines with Mixture Kernels

Wei

Wan

2013

AMR

View full text Add to dashboard Cite

Recent studies have showed that machine learning techniques are advantageous to statistical models for medicine database classification, such as SVM. In this study, we discuss the applications of the support vector machine with mixture of kernel (SVM-MK) to design a classification system. Differing from the standard SVM, the SVM-MK uses the 1-norm based object function and adopts the convex combinations of single feature basic kernels. Only a linear programming problem needs to be resolved and it greatly reduces the computational costs. More important, it is a transparent model and the optimal feature subset can be obtained automatically. Three UCI databases are used to demonstrate the good performance of the SVM- MK.

show abstract

Section: Introductionmentioning

confidence: 99%

Data Classification Using Support Vector Machines with Mixture Kernels

Wei

Wan

2013

AMR

View full text Add to dashboard Cite

show abstract

“…Support vector machines (SVMs) [33] are a class of state-of-the-art classification algorithm that is known to be successful in a wide variety of applications because of their strong robustness, high generalization ability, and low classification error. However, an obvious drawback is that a SVM lacks explanation capability for their results when applied to classification problems, that is, the results obtained from SVM classifiers are not intuitive to humans and are hard to understand [34] . For example, when an unlabeled example is classified by the SVM classifier as positive or negative, the only explanation that can be provided is that the outputs of the variables are lower/higher than some threshold; such an explanation is completely non-intuitive to human experts [34] .…”

mentioning

confidence: 99%

“…However, an obvious drawback is that a SVM lacks explanation capability for their results when applied to classification problems, that is, the results obtained from SVM classifiers are not intuitive to humans and are hard to understand [34] . For example, when an unlabeled example is classified by the SVM classifier as positive or negative, the only explanation that can be provided is that the outputs of the variables are lower/higher than some threshold; such an explanation is completely non-intuitive to human experts [34] . Therefore, it is often difficult to extract rules from a trained SVM classifier.…”

mentioning

confidence: 99%

Designing a Hybrid Intelligent Mining System for Credit Risk Evaluation

Wang

Wen

et al. 2008

J Syst Sci Complex

View full text Add to dashboard Cite

In this study, a novel hybrid intelligent mining system integrating rough sets theory and support vector machines is developed to extract efficiently association rules from original information table for credit risk evaluation and analysis. In the proposed hybrid intelligent system, support vector machines are used as a tool to extract typical features and filter its noise, which are different from the previous studies where rough sets were only used as a preprocessor for support vector machines. Such an approach could reduce the information table and generate the final knowledge from the reduced information table by rough sets. Therefore, the proposed hybrid intelligent system overcomes the difficulty of extracting rules from a trained support vector machine classifier and possesses the robustness which is lacking for rough-set-based approaches. In addition, the effectiveness of the proposed hybrid intelligent system is illustrated with two real-world credit datasets.

show abstract

“…Existem diversos trabalhos que abordam esse tema (Fu, 1991;Hayashi, 1991;Towell and Shavlik, 1993;Craven and Shavlik, 1994b;Fu, 1994;Sethi and Yoo, 1994;Tan, 1994;Thrun, 1994;Tickle et al, 1994;Alexander and Dietterich, 1995;Setiono and Liu, 1995;Thrun, 1995;Craven, 1996;Milaré et al, 1997;Schellharmmer et al, 1997;Tickle et al, 1998;Martineli, 1999;Schmitz et al, 1999;Duch et al, 2000;Behloul et al, 2002;Zhou and Jiang, 2002;Milaré et al, 2002;Núñez et al, 2002;Fung et al, 2008).…”

Section: Conjunto Satimage -mentioning

confidence: 99%

“…Esse trabalho, proposto por Fung et al (2008) descreve um método para a conversão de SVMs lineares em um conjunto de regras sem sobreposição. Nesse método, cada iteração do algoritmo de extração de regrasé formulada como um problema de otimização com restrições.…”

Section: Extração De Conhecimento De Svmsunclassified

Extração de conhecimento simbólico em técnicas de aprendizado de máquina caixa-preta por similaridade de rankings

Bianchi¹

View full text Add to dashboard Cite

Non-symbolic Machine Learning techniques, like Artificial Neural Networks, Support Vector Machines and Ensembles of classifiers have shown a good performance when they are used in data analysis. The strong limitation regarding the use of these techniques is the lack of comprehensibility of the knowledge stored in their internal structure. This Thesis presents an investigation of methods capable of extracting comprehensible representations of the knowledge acquired by these non-symbolic techniques, here named black box, during their learning process. The main contribution of this work is the proposal of a new pedagogical method for rule extraction that explains the classification process followed by non-symbolic techniques. This new method is based on the optimization (maximization) of the similarity between classification rankings produced by symbolic and non-symbolic (from where the internal knowledge is being extracted) Machine Learning techniques. Experiments were performed for several datasets and the results obtained suggest a good potential of the proposed method.

show abstract

Rule Extraction from Linear Support Vector Machines via Mathematical Programming

Cited by 7 publications

References 12 publications

Data Classification Using Support Vector Machines with Mixture Kernels

Data Classification Using Support Vector Machines with Mixture Kernels

Designing a Hybrid Intelligent Mining System for Credit Risk Evaluation

Extração de conhecimento simbólico em técnicas de aprendizado de máquina caixa-preta por similaridade de rankings

Contact Info

Product

Resources

About