Multiclass SVM Model Selection Using Particle Swarm Optimization

Souza, Bruno de Castro e; Carvalho, André C. P. L. F. de; Calvo, Rodrigo; Ishii, Renato Porfírio

doi:10.1109/his.2006.264914

Cited by 43 publications

(36 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Still, some comparisons can be made. Souza et al [6] report a classification accuracy of 0.9882 using k-fold validation on the training set for the Vowel dataset. KGP returns an average accuracy of 0.9981 on the training set.…”

Section: Experiments and Resultsmentioning

confidence: 99%

“…Phienthrakul and Kilsirikul [20] used ES to learn the weights in a weighted linear combination of Gaussian radial basis functions. Souza et al [6] use Particle Swarm Optimization (PSO) to learn optimal parameters in a Gaussian kernel function for multi-class classification, while Huang and Wang use a GA for the same task [13,12]. Runarsson and Sigurdsson [21] use a parallel ES to learn optimal parameters in a Gaussian kernel.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Evolving kernels for support vector machine classification

Sullivan

Luke

2007

Proceedings of the 9th Annual Conference on Genetic and Evolutionary Computation

View full text Add to dashboard Cite

While support vector machines (SVMs) have shown great promise in supervised classification problems, researchers have had to rely on expert domain knowledge when choosing the SVM's kernel function. This project seeks to replace this expert with a genetic programming (GP) system. Using strongly typed genetic programming and principled kernel closure properties, we introduce a new algorithm, called KGP, which finds near-optimal kernels. The algorithm shows wide applicability, but the combined computational overhead of GP and SVMs remains a major unresolved issue.

show abstract

Section: Experiments and Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Evolving kernels for support vector machine classification

Sullivan

Luke

2007

Proceedings of the 9th Annual Conference on Genetic and Evolutionary Computation

View full text Add to dashboard Cite

show abstract

“…It can be applied to other parametric classifiers (KNN, Neural network, etc.) with other optimization methods ( [9]). Moreover, it can be easily extended through the introduction of other parameters (kernel type) or objectives (number of support vectors, decision time).…”

Section: Discussionmentioning

confidence: 99%

Noncost Sensitive SVM Training Using Multiple Model Selection

Chatelain

Adam

Lecourtier

et al. 2010

J CIRCUIT SYST COMP

View full text Add to dashboard Cite

Abstract. In this paper, we propose a multi-objective optimization framework for SVM hyperparameters tuning. The key idea is to manage a population of classifiers optimizing both False Positive (FP) and True Positive (TP) rates rather than a single classifier optimizing a scalar criterion. Hence, each classifier in teh population optimizes a particular trade-off between the objectives. Within the context of two-class classification problems, our work introduces the "ROC front concept" depicting a population of SVM classifiers as an alternative to the ROC curve representation. The comparison with a traditional scalar optimization technique based on an AUC criterion shows promising results on UCI datasets.

show abstract

“…In order to automatize this process and to avoid an exhaustive or a random exploration of parameters, different authors have deployed search and optimization techniques [3,4,6,7,8,13,14,15]. In this context, the search space consists on a set of possible configurations of parameters and the objective function corresponds to a performance measure (e.g., precision estimated by cross-validation) obtained by the SVM on the problem.…”

Section: Svm Parameter Selectionmentioning

confidence: 99%