Multiclass SVM-RFE for product form feature selection

Shieh, Meng-Dar; Yang, Chulho

doi:10.1016/j.eswa.2007.07.043

Cited by 81 publications

(32 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…SVM-RFE has been extended for variable selection in multiclass classification (Zhou and Tuck, 2007;Zhao and Yand, 2010;Shieh and Yang, 2008;Duan et al, 2007). To deal with multiple classes, multi-class problems can be decomposed into several binary classification problems (Zhou and Tuck, 2007;Shieh and Yang, 2008;Duan et al, 2007). Assuming that all classes equally contribute to the classification, the variable that simultaneously minimizes all of the variable selection criteria of binary classification problems is removed.…”

Section: Related Studiesmentioning

confidence: 99%

“…The tested methods are presented in Table 3. The variable selection methods are two multi-class feature scoring methods, namely Chi-squared (CHI) (Yang and Pedersen, 1997;Forman, 2003) and Information Gain (IG) (Yang and Pedersen, 1997;Forman, 2003;Quinlan, 1993), along with the multi-class support vector machine-recursive feature elimination (MSVM-RFE) (Zhou and Tuck, 2007;Shieh and Yang, 2008).…”

Section: Computational Settingmentioning

confidence: 99%

“…Assuming that all classes equally contribute to the classification, the variable that simultaneously minimizes all of the variable selection criteria of binary classification problems is removed. In this paper, as the variable selection criterion, we use the summation of the variable selection criteria of the one-against-all SVM using a Gaussian kernel (Zhou and Tuck, 2007;Shieh and Yang, 2008).…”

Section: Related Studiesmentioning

confidence: 99%

“…Among such embedded methods, some select different variable subsets for each binary classifier using a variable selection method for binary classification (Veenman and Bolck, 2011;Ramaswamy et al, 2001;Chai and Domeniconi, 2004). Other methods compute the selection criteria values of all of the variables for each of the binary classifiers and select, by combining those values, a common set of variables for all of the binary classifiers (Chen et al, 2006;Duan et al, 2007;Chapelle and Keerthi, 2008;Liu et al, 2008;Shieh and Yang, 2008;Zhou and Tuck, 2007).…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Variable selection methods for multi-class classification using signomial function

Hwang

Lee

Park

2017

Journal of the Operational Research Society

View full text Add to dashboard Cite

We develop several variable selection methods using signomial function to select relevant variables for multiclass classification by taking all classes into consideration. We introduce a ' 1 -norm regularization function to measure the number of selected variables and two adaptive parameters to apply different importance weights for different variables according to their relative importance. The proposed methods select variables suitable for predicting the output and automatically determine the number of variables to be selected. Then, with the selected variables, they naturally obtain the resulting classifiers without an additional classification process. The classifiers obtained by the proposed methods yield competitive or better classification accuracy levels than those by the existing methods.

show abstract

Section: Related Studiesmentioning

confidence: 99%

Section: Computational Settingmentioning

confidence: 99%

Section: Related Studiesmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Variable selection methods for multi-class classification using signomial function

Hwang

Lee

Park

2017

Journal of the Operational Research Society

View full text Add to dashboard Cite

show abstract

“…Recursive feature elimination for support vector machine (RFE-SVM) [16] is such a feature selection algorithm. It was originally formulated for binary classification problems, and then extended for multi-class problems [10,23] and some other variants were also proposed [15,17,21]. The goal of RFE-SVM is to find a subset of m from M candidate features, m oM, which maximizes the performance of an SVM classifier.…”

Section: Brief State-of-the-art Reviewmentioning

confidence: 99%

An efficient feature selection method for mobile devices with application to activity recognition

et al. 2011

View full text Add to dashboard Cite

Application of Evolutionary Neural Networks on Optimization Design of Mobile Phone Based on User's Emotional Needs

Guo

Peng

et al. 2015

Hum Ftrs & Erg Mfg Svc

View full text Add to dashboard Cite

Taking users' emotional needs into consideration, this research aims to propose a new method to present product design features exactly and completely. On the basis of genetic algorithm integrated with backpropagation (BP) neural networks, taking the mobile phone as research object, an optimization design algorithm was finally designed. First, the continuous and discrete design variables that describe mobile phones were screened with methods of dimensions, coordinate label, and morphological analysis. Forty three-dimensional (3D) mobile phone models were designed by using 3D design software PROE. Accordingly, 12 representative mobile phones were selected through multidimensional scaling analysis and cluster analysis. Fourteen pairwise Kansei image words were obtained by collecting, screening, surveys, and statistical analysis method. Second, a BP neural networks model between design variables and user preference along with Kansei image words was established and verified with questionnaire survey data. Finally, the optimization design model for mobile phones was established considering design requirements and users' emotional needs. A genetic algorithm integrated with BP neural networks was used to optimize mobile phone design. The results show that the optimization scheme is superior to others, and this paper will provide design suggestion for mobile phone designers. C 2015 Wiley Periodicals, Inc.

show abstract

Multiclass SVM-RFE for product form feature selection

Cited by 81 publications

References 25 publications

Variable selection methods for multi-class classification using signomial function

Variable selection methods for multi-class classification using signomial function

An efficient feature selection method for mobile devices with application to activity recognition

Application of Evolutionary Neural Networks on Optimization Design of Mobile Phone Based on User's Emotional Needs

Contact Info

Product

Resources

About