Dimensionality Reduction on Text via Feature Selection

Feature selection, an effective technique for dimensionality reduction, plays an important role in many machine learning systems. Supervised knowledge can significantly improve the performance. However, faced with the rapid growth of newly emerging concepts, existing supervised methods might easily suffer from the scarcity and validity of labeled data for training. In this paper, the authors study the problem of zero-shot feature selection (i.e., building a feature selection model that generalizes well to “unseen” concepts with limited training data of “seen” concepts). Specifically, they adopt class-semantic descriptions (i.e., attributes) as supervision for feature selection, so as to utilize the supervised knowledge transferred from the seen concepts. For more reliable discriminative features, they further propose the center-characteristic loss which encourages the selected features to capture the central characteristics of seen concepts. Extensive experiments conducted on various real-world datasets demonstrate the effectiveness of the method.

Download Full-text

Feature Selection in Pathology Detection using Hybrid Multidimensional Analysis

Encyclopedia of Healthcare Information Systems ◽

10.4018/978-1-59904-889-5.ch074 ◽

2008 ◽

pp. 587-593 ◽

Cited By ~ 2

Author(s):

Edilson Delgado-Trejos ◽

Germán Castellanos ◽

Luis G. Sánchez ◽

Julio F. Suárez

Keyword(s):

Feature Selection ◽

Multivariate Analysis ◽

Variable Selection ◽

Dimensionality Reduction ◽

Cleft Lip ◽

Feature Selection Method ◽

Bivariate Analysis ◽

Quality Level ◽

Relevant Variables ◽

Complex Features

Dimensionality reduction procedures perform well on sets of correlated features while variable selection methods perform poorly. These methods fail to pick relevant variables because the score they assign to correlated features is too similar, and none of the variables is strongly preferred over another. Hence, feature selection and dimensionality reduction algorithms have complementary advantages and disadvantages. Dimensionality reduction algorithms thrive on the correlation between variables but fail to select informative features from a set of more complex features. Variable selection algorithms fail when all the features are correlated, but succeed with informative variables (Wolf & Bileschi, 2005). In this work, we propose a feature selection algorithm with heuristic search that uses Multivariate Analysis of Variance (MANOVA) as the cost function. This technique is put to the test by classifying hypernasal from normal voices of CLP (cleft lip and/or palate) patients. The classification performance, computational time, and reduction ratio are also considered by comparing with an alternate feature selection method based on the unfolding of multivariate analysis into univariate and bivariate analysis. The methodology is effective because it has in mind the statistical and geometrical relevance present in the features, which does not summarize the analysis of the separability among classes, but searches a quality level in signal representation.

Download Full-text

Dimensionality Reduction: Is Feature Selection More Effective Than Random Selection?

10.1007/978-3-030-85030-2_10 ◽

2021 ◽

pp. 113-125

Author(s):

Laura Morán-Fernández ◽

Verónica Bolón-Canedo

Keyword(s):

Feature Selection ◽

Dimensionality Reduction ◽

Random Selection

Download Full-text

Research on the Feature Selection of Rolling Bearings’ Degradation Features

Shock and Vibration ◽

10.1155/2019/6450719 ◽

2019 ◽

Vol 2019 ◽

pp. 1-19

Author(s):

Yaolong Li ◽

Hongru Li ◽

Bing Wang ◽

He Yu ◽

Weiguo Wang

Keyword(s):

Feature Selection ◽

Dimensionality Reduction ◽

Root Mean Square ◽

Sample Entropy ◽

Remaining Useful Life ◽

Mean Square ◽

Rolling Bearings ◽

Reduction Methods ◽

Useful Life ◽

Selection Of

The bearings’ degradation features are crucial to assess the performance degradation and predict the remaining useful life of rolling bearings. So far, numerous degradation features have been proposed. Many researchers have devoted to use dimensionality reduction methods to reduce the redundancy of those features. However, they have not considered the properties and similarity of those features. In this paper, we present a simple way to reduce dimensionality by classifying different features based on their trends. And the degradation features can be classified into two subdivisions, namely, uptrends and downtrends. In each subdivision, there exists visible trend similarity, and we have introduced two indexes to measure this similarity. By selecting the representative features of the subdivision, the multifeatures can be dimensionality reduced. Through the comparison, the root mean square and sample entropy are two good representatives of uptrend and downtrend features. This method gives an alternative way for dimensionality reduction of the rolling bearings’ degradation features.

Download Full-text