SVM-FuzCoC: A novel SVM-based feature selection method using a fuzzy complementary criterion

Moustakidis, Serafeim; Theocharis, John B.

doi:10.1016/j.patcog.2010.05.007

Cited by 47 publications

(29 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The water-absorption bands were discarded resulting in 191 bands, where 420 locations were selected and labeled into seven classes (wdcBands). Because this is an urban classification task, structural features (SF) can effectively improve the classification performance, as it has been proven by previous 47 . The new dataset (wdcBandsSF) comprises a total of 281 features (191 bands + 6×15 SF).…”

Section: Comparative Analysismentioning

confidence: 88%

Reducing the Complexity of Genetic Fuzzy Classifiers in Highly-Dimensional Classification Problems

Stavrakoudis¹,

Galidaki²,

Gitas³

et al. 2012

IJCIS

View full text Add to dashboard Cite

This paper introduces the Fast Iterative Rule-based Linguistic Classifier (FaIRLiC), a Genetic Fuzzy Rule-Based Classification System (GFRBCS) which targets at reducing the structural complexity of the resulting rule base, as well as its learning algorithm's computational requirements, especially when dealing with high-dimensional feature spaces. The proposed methodology follows the principles of the iterative rule learning (IRL) approach, whereby a rule extraction algorithm (REA) is invoked in an iterative fashion, producing one fuzzy rule at a time. The REA is performed in two successive steps: the first one selects the relevant features of the currently extracted rule, whereas the second one decides the antecedent part of the fuzzy rule, using the previously selected subset of features. The performance of the classifier is finally optimized through a genetic tuning post-processing stage. Comparative results in a hyperspectral remote sensing classification as well as in 12 real-world classification datasets indicate the effectiveness of the proposed methodology in generating high-performing and compact fuzzy rule-based classifiers, even for very high-dimensional feature spaces.

show abstract

Section: Comparative Analysismentioning

confidence: 88%

Reducing the Complexity of Genetic Fuzzy Classifiers in Highly-Dimensional Classification Problems

Stavrakoudis¹,

Galidaki²,

Gitas³

et al. 2012

IJCIS

View full text Add to dashboard Cite

show abstract

“…Researchers have proposed many single-view feature selection algorithms [16,25,27], which are classified into three kinds: Wrappers, Filters and Embedded methods [11]. The Wrapper methods evaluate the importance of features by estimating the performance of learned classifiers; the Filter methods select features according to several correlation criteria; the Embedding methods have an one-step learning process by combining feature selection and classification.…”

Section: Single-view Feature Selectionmentioning

confidence: 99%

“…Specifically, in many real problems, the features from different views sometimes are redundant, irrelevant or even noisy. Traditional feature selection methods [16,25,27] have been developed to select the relevant features and eliminate the redundant features from a single feature space by a specific selection metric [13] and an efficient searching strategy [33] (See Section 2.2). However, they cannot be directly borrowed for multi-view data, since the multi-view data has multiple connected but different feature spaces.…”

Section: Introductionmentioning

confidence: 99%

mPadal: a joint local-and-global multi-view feature selection method for activity recognition

et al. 2014

View full text Add to dashboard Cite

The selection of multi-view features plays an important role for classifying multi-view data, especially the data with high dimension. In this paper, a novel multi-view feature selection method via joint local patterndiscrimination and global label-relevance analysis (mPadal) is proposed. Different from the previous methods which globally select the multi-view features directly via viewlevel analysis, the proposed mPadal employs a new joint local-and-global way. In the local selection phase, the pattern-discriminative features will be first selected by considering the local neighbor structure of the most discriminative patterns. In the global selection phase, the features with the topmost label-relevance, which can well separate different classes in the current view, are selected. Finally, the two parts selected are combined to form the final features. Experimental results show that compared with several baseline methods in publicly available activity recognition dataset IXMAS, mPadal performs the best in terms of the highest accuracy, precision, recall and F1 score. Moreover, the features selected by mPadal are highly complementary among views for classification, which is able to improve the classification performance according to previous theoretical studies.

show abstract

“…An efficient feature selection technique will be employed to reduce the highdimensionality of the recorded ultrasonic signal and provide a more descriptive and compact signal representation [10]. In addition to enhanced accuracy rates and dimensionality reduction, the method has reasonably low computational demands, being able to cope with the high-dimensional feature space.…”

Section: Feature Selectionmentioning

confidence: 99%

An automated long range ultrasonic rail flaw detection system based on the support vector machine algorithm

Moustakidis

Kappatos

Karlsson

et al. 2012

WIT Transactions on the Built Environment

View full text Add to dashboard Cite

A contact force between pantograph and contact line is one of the important criteria to evaluate contact performance. Hence, contact force measuring methods were developed. The currently-used measurement methods use built-in sensors on the panhead. However, these sensors affected the dynamic characteristics of the pantograph. In addition, the methods need data transmitting equipment, because built-in sensors are set under a high-voltage condition. To eliminate these drawbacks, the authors have developed a new contact force measuring method which can replace the all built-in sensors with image processing sensors mounted on the rooftop of the train. To verify measurement accuracy of this method, we performed an excitation test of the pantograph. Experimental results show that the measurement method meets the requirements of the EN standard.

show abstract

SVM-FuzCoC: A novel SVM-based feature selection method using a fuzzy complementary criterion

Cited by 47 publications

References 33 publications

Reducing the Complexity of Genetic Fuzzy Classifiers in Highly-Dimensional Classification Problems

Reducing the Complexity of Genetic Fuzzy Classifiers in Highly-Dimensional Classification Problems

mPadal: a joint local-and-global multi-view feature selection method for activity recognition

An automated long range ultrasonic rail flaw detection system based on the support vector machine algorithm

Contact Info

Product

Resources

About