Feature Selection and Granularity Learning in Genetic Fuzzy Rule-Based Classification Systems for Highly Imbalanced Data-Sets

Villar, Pedro; Fernández, Alberto; Carrasco, R.A.; Herrera, F.

doi:10.1142/s0218488512500195

Cited by 32 publications

(32 citation statements)

References 52 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In [90], a genetic procedure for learning the KB in imbalanced datasets, GA-FS+GL, is proposed. In this case, the SMOTE algorithm is again used to balance the training set.…”

Section: Efs and Data-level Approachesmentioning

confidence: 99%

“…Finally, in [110], three EFS systems for imbalanced classification are compared, which are the GA-FS+GL method described in [90], the GP-COACH-H algorithm presented in [19] and the MOEFS developed in [107]. The authors use 22 datasets to perform this comparison being the MOEFS approach the one with the best performance supported by a Holm test.…”

Section: Efs and Ensemble Learningmentioning

confidence: 99%

See 1 more Smart Citation

Evolutionary Fuzzy Systems: A Case Study in Imbalanced Classification

Fernández

Herrera

2016

Fuzzy Logic and Information Fusion

View full text Add to dashboard Cite

The use of evolutionary algorithms for designing fuzzy systems provides them with learning and adaptation capabilities, resulting on what is known as Evolutionary Fuzzy Systems. These types of systems have been successfully applied in several areas of Data Mining, including standard classification, regression problems and frequent pattern mining. This is due to their ability to adapt their working procedure independently of the context we are addressing. Specifically, Evolutionary Fuzzy Systems have been lately applied to a new classification problem showing good and accurate results. We are referring to the problem of classification with imbalanced datasets, which is basically defined by an uneven distribution between the instances of the classes. In this work, we will first introduce some basic concepts on linguistic fuzzy rule based systems. Then, we will present a complete taxonomy for Evolutionary Fuzzy Systems. Then, we will review several significant proposals made in this research area that have been developed for addressing classification with imbalanced datasets. Finally, we will show a case study from which we will highlight the good behavior of Evolutionary Fuzzy Systems in this particular context.

show abstract

“…In [90], a genetic procedure for learning the KB in imbalanced datasets, GA-FS+GL, is proposed. In this case, the SMOTE algorithm is again used to balance the training set.…”

Section: Efs and Data-level Approachesmentioning

confidence: 99%

Section: Efs and Ensemble Learningmentioning

confidence: 99%

Evolutionary Fuzzy Systems: A Case Study in Imbalanced Classification

Fernández

Herrera

2016

Fuzzy Logic and Information Fusion

View full text Add to dashboard Cite

show abstract

“…Several papers, mainly related to MOEFSs, adopt multiobjective evolutionary algorithms for rule selection (Gacto et al 2010) (e.g., from an initial RB heuristically generated) or rule learning (Cococcioni et al 2007), DB tuning (Botta et al 2009), rule learning/selection together with DB learning (in particular, partition granularity and membership function parameters) (Villar et al 2012;Antonelli et al 2009a, b). As regards other formalisms for representing information granules, in the last years, a number of evolutionary-based approaches, mainly based on singleobjective optimization, have been proposed (Castillo and Melin 2012a, b;Sanz et al 2010Sanz et al , 2011Sanz et al , 2013Sanz et al , 2015.…”

Section: Introductionmentioning

confidence: 99%

Multi-objective evolutionary design of granular rule-based classifiers

et al. 2015

View full text Add to dashboard Cite

In the last years, rule-based systems have been widely employed in several different application domains. The performance of these systems is strongly affected by the process of information granulation, which defines in terms of specific information granules such as sets, fuzzy sets and rough sets, the labels used in the rules. Generally, information granules are either provided by an expert, when possible, or extracted from the available data. In the framework of rule-based classifiers, we investigate the importance of determining an effective information granulation from data, preserving the comprehensibility of the granules. We show how the accuracies of rule-based classifiers can be increased by learning number and parameters of the granules, which partition the involved variables. To perform this analysis, we exploit a multi-objective evolutionary approach to the classifier generation we have recently proposed. We discuss different levels of information granulation optimization employing both the learning of the number of granules per variable and the tuning of each granule during the evolutionary process. We show and discuss the results obtained on several classification benchmark datasets using fuzzy sets and intervals as types of information granules.

show abstract

“…It is well known that a large number of features can degrade the discovery of the borderline areas of the problem [39], either because some of these variables might be redundant or because they do not show a good synergy among them. For this reason, some works on the topic have proposed the use of feature selection for imbalanced datasets in order to overcome this problem [54], and to diminish the effect of overlapping [13]. However, the application of feature selection might be too aggressive and therefore some potential features could be discarded.…”

Section: Introductionmentioning

confidence: 99%