A hybrid feature selection method for DNA microarray data

A Survey on Filter Techniques for Feature Selection in Text Mining

“…Chuang et al [4] combine a filter FS method correlation-based feature selection (CFS) and a wrapper FS method taguchi-genetic algorithm (TGA) to form a new hybrid method for dimension reduction in gene analysis. They use k-nearest neighbor (KNN) with leave-one-out cross-validation (LOOCV) method as a classifier to judge the efficacy of their proposed method and observe that their proposed method …”

Distributed feature selection: An application to microarray data classification

“…Wang et al [20] also proposed a novel filter framework to select optimal feature subsets based on a maximum weight and minimum redundancy criterion. Hybrid methods have been recently tested on this type of data [21,22] obtaining high classification accuracies. Embedded methods have also been proposed, such as in [23], where the authors introduced an algorithm that simultaneously selects relevant features during classifier construction by penalizing each feature's use in the dual formulation of support vector machines.…”

“…Table 12 displays the best results for our proposed distributed approaches (rows 1-15), two state of the art (SofA) methods; the MWMR method proposed in [20] and an ensemble of filters proposed in [6] (rows 16 and 17), five filters widely applied to microarray data in a centralized fashion (rows [18][19][20][21][22]; CFS, consistency-based, INTERACT, Information Gain and ReliefF, respectively) and the four classifiers considered in this study when no feature selection is applied (rows 23-26). Note that the last column reports the average for all the microarray datasets and that Breast dataset could not be compared because it was not included in the experiments carried out in [7].…”

Feature Selection in DNA Microarray Classification

“…Chuang et al [62] proposed a hybrid method called CFS-TGA, which combines correlation-based feature selection (CFS) and the Taguchi-genetic algorithm, where the k-nearest neighbor served as a classifier. The proposed method obtained the highest classification accuracy in ten out the 11 gene expression datasets it was tested on.…”

“…k-fold cross-validation is a common choice [42,44,48,57,58,61,68], as is holdout validation [43,46,52,55,59,63,69]. Bootstrap sampling was used less [50,60,66], probably due to its high computational cost, and there are also some representatives of leave-one-out cross-validation [62,64].…”