Search

• A systematic evaluation of filter Unsupervised Feature Selection methods is presented. • The most popular and recent filter UFS methods are included in our study. • The evaluation of the filter UFS methods followed the standards in the literature. • A general discussion based on the results of the evaluated methods is provided. • Some guidelines for the use of the evaluated filter UFS methods is also provided. Unsupervised Feature Selection (UFS) has aroused great interest in the last years because of its practical significance and application on a large variety of problems in expert and intelligent systems where unlabeled data appear. Specifically, Unsupervised Feature Selection methods based on the filter approach have received more attention due to their efficiency, scalability, and simplicity. However, in the literature, there are no comprehensive studies for assessing such UFS methods when they are applied, under the same conditions, to a wide variety of real-world data. To fill this gap, in this paper, we present a comprehensive empirical and systematic evaluation of the most popular and recent filter UFS methods, evaluating their performance in terms of clustering, classification, and runtime. The filter methods used in our study were applied on 50 datasets from the UCI Machine Learning Repository and 25 high dimensional datasets from the ASU Feature Selection Repository. To evaluate if the outcomes obtained by the assessed methods are statistically significant, the Friedman test and Holm post hoc procedure were applied in the clustering and classification results. From our experiments, we provide some practical guidelines and insights for the use of the filter UFS methods analyzed in our study. [ABSTRACT FROM AUTHOR]

Abstract: In pattern recognition, feature selection is a very important task for supervised classification. The problem consists in, given a dataset where each object is described by a set of features, finding a subset of the original features such that a classifier that runs on data containing only these features would reach high classification accuracy. A useful way to find this subset of the original features is through testor theory. A testor is defined as a subset of the original features that allows differentiating objects from different classes. Testors are very useful particularly when object descriptions contain both numeric and non-numeric features. Computing testors for feature selection is a very complex problem due to exponential complexity, with respect to the number of features, of algorithms based on testor theory. Hardware implementation of testor computing algorithms helps to improve their performance taking advantage of parallel processing for verifying if a feature subset is a testor in a single clock cycle. This paper introduces an efficient hardware–software platform for computing irreducible testors for feature selection in pattern recognition. Results of implementing the proposed platform using a FPGA-based prototyping board are presented and discussed. [Copyright &y& Elsevier]