Hybridization of Evolutionary Mechanisms for Feature Subset Selection in Unsupervised Learning

Torres, Dolores; Ponce-de-Leon, Eunice; Torres, Aurora; Ochoa, Alberto; Díaz, Elizabeth

doi:10.1007/978-3-642-05258-3_54

Cited by 6 publications

(12 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Please note that we do not make comparisons with the GA published in [15], because the authors did not provide the algorithm to make comparisons with the proposal Hill Climbing algorithm. Besides, the only dataset presented in [15], has 29 features, which is regarded as relatively small Basic Matrix.…”

Section: Remarkmentioning

confidence: 98%

“…Besides, the only dataset presented in [15], has 29 features, which is regarded as relatively small Basic Matrix. The number of rows of Basic Matrix above is great.…”

Section: Remarkmentioning

confidence: 99%

“…In general, two approaches have been developed to address this problem: a) generate the whole set of typical testors (LEX [13], fast-CT EXT [11]); and b) find a subset of typical testors (GA [12], UMDA [1], AGHPIA [15]). Nevertheless, these global search heuristics become too slow as the number of features grows because the goal of these techniques focuses on finding the entire typical-testor set as global optimum.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Typical Testors Generation Based on an Evolutionary Algorithm

Diaz-Sanchez

Piza-Davila

Sánchez-Díaz

et al. 2011

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. Typical testors are useful for both feature selection and feature relevance determination in supervised classification problems. However, reported algorithms that address the problem of finding the set of all typical testors have exponential complexity. In this paper, we propose to adapt an evolutionary method, the Hill-Climbing algorithm, with an acceleration operator in mutation process, to address this problem in polinomial time. Experimental results with the method proposed are presented and compared, in efficiency, with other methods, namely, Genetic Algorithms (GA) and Univariate Marginal Distribution Algorithm (UMDA).

show abstract

Section: Remarkmentioning

confidence: 98%

“…Besides, the only dataset presented in [15], has 29 features, which is regarded as relatively small Basic Matrix. The number of rows of Basic Matrix above is great.…”

Section: Remarkmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Typical Testors Generation Based on an Evolutionary Algorithm

Diaz-Sanchez

Piza-Davila

Sánchez-Díaz

et al. 2011

Lecture Notes in Computer Science

View full text Add to dashboard Cite

show abstract

“…Regularly, Featured Subset Selection (FSS) [13] is used to reduce dimensionality [14], which is used to efficiently reduce the number of variables, attributes or characteristics with which should describe the objects and to find their influence in a problem. This is an alternative method that starts by using the set of typical testors, taking out irrelevant or redundant features [11,14].…”

Section: Featured Subset Selectionmentioning

confidence: 99%

“…FSS really has importance because reducing the number of features may help to decrease the cost of acquiring data and also make the classification models easier to understand [14,15]. Also, the number of features could affect the accuracy of classification.…”

Section: Featured Subset Selectionmentioning

confidence: 99%

Feature Subset Selection and Typical Testors Applied to Breast Cancer Cells

Gallegos¹,

Torres²,

Álvarez³

et al. 2016

RCS

Self Cite

View full text Add to dashboard Cite

One of the most significant concerns around the world has been human health. So far, there have been little discussion about how the Computer Science can help to improve diagnostics of many diseases. In this paper, the aim is to prove that the Computer Science can offer techniques to help the improvement of the diagnosis of medical pathologies. For that, Feature Subset Selection and Typical Testors will be applied to a breast cancer database. Nowadays, cancer is a medical condition difficult to ignore, which has special interest by the specialists for finding effective methods to prevent and cure it. The database contains a feature set of breast cancer cells, which were subjected to an analysis of testors in order to find the minimum feature set that best describes benign and malignant cells. As a result, each typical testor found contains risk factors recognized by medical experts.

show abstract

Identification of Risk Factors for TRALI Using a Hybrid Algorithm

Torres

Cuellar³

et al. 2013

Advances in Artificial Intelligence

View full text Add to dashboard Cite

Hybridization of Evolutionary Mechanisms for Feature Subset Selection in Unsupervised Learning

Cited by 6 publications

References 18 publications

Typical Testors Generation Based on an Evolutionary Algorithm

Typical Testors Generation Based on an Evolutionary Algorithm

Feature Subset Selection and Typical Testors Applied to Breast Cancer Cells

Identification of Risk Factors for TRALI Using a Hybrid Algorithm

Contact Info

Product

Resources

About