A fast feature selection approach based on extreme learning machine and coefficient of variation

Abstract

Abstract:Feature selection is the method of reducing the size of data without degrading their accuracy. In this study, we propose a novel feature selection approach, based on extreme learning machines (ELMs) and the coefficient of variation (CV). In the proposed approach, the most relevant features are identified by ranking each feature with the coefficient obtained through ELM divided by CV. The achieved accuracies and computational costs, obtained with the use of features selected via the proposed approach in 9 classification and 26 regression benchmark data sets, were compared to those obtained with all features, as well as those obtained with the features selected by a wrapper and a filtering method. The achieved accuracy values obtained with the proposed approach were generally higher than when using all features. Furthermore, high feature reduction ratios were obtained with the proposed approach, including the achieved feature reduction ratios in epilepsy, liver, EMG, shuttle, and abalone. Stock data sets were 90.48%, 90%, 70.59%, 66.67%, 75%, and 77.78%, respectively. This approach is an extremely fast process that is independent of the employed machine-learning methods.

On a yearly basis prediction of soil water content utilizing sar data: a machine learning and feature selection approach

“…The feature selection method is one of the significant issues in the areas of machine learning and it is generally used in the pattern detection and recognition applications [32]. The ELM, which has a high-speed training phase and good generalization capacity, is considered a fine approach for the feature selection method [42][43][44].…”

“…Although ELM is generally used as a machine learning method, it has been developed by [32] as a consistent, fast and independent feature selection algorithm with better accuracy [32]. In this method, biases and weights of the hidden layer are allocated indiscriminately, and the output layer's weights are computed with the aid of the Moore-Penrose approach [45,46].The output of ELM, which has one single neuron, can be determined in Equation 3.…”

“…However, there are still several problems in the above methods such as the inadequacy of relations between the features, the complexity of high computation, monotonicity of objective function [31]. Therefore, a new approach using extreme learning based feature selection (ELM-FS) is preferred in this study since it has fast learning speed and good generalization performance comparing to the other methods [32].…”

Detection of Covid-19 from X-ray Images via Ensemble of Features Extraction Methods Employing Randomized Neural Networks

“…Therefore, these parameters can be chosen as random, and the output weights can be analytically ascertained. This situation allows RNN to be easy and fast in data processing [32] and ensures a fine generalization achievement for a single FFNN [33]. Compared to the other known gradient-based learning algorithms, RNN has several advantages like the potential of reaching the minimum training error, operating with non-differentiable activation functions, and employing a single hidden layer [34].…”