A Novel Unsupervised Feature Selection Approach Using Genetic Algorithm on Partitioned Data

Aug 2, 2026

Uploaded by Rowan

A novel unsupervised feature selection method is introduced to reduce high-dimensional data while preserving representative structure. Sammon’s Stress Function maps data into a lower-dimensional space, then the dataset is partitioned and features are assigned to partitions randomly. A genetic algorithm uses Sammon error as fitness to select a target number of features per partition, and the reduced feature subsets are iteratively repackaged for further refinement. Experiments on 11 standard datasets with Decision Tree, MLP, and KNN show higher accuracy on most datasets and fewer selected features than prior methods.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages16

Reading Time40 min

AccessFree to download

Abstract

Introduction

Motivation: curse of dimensionality

Feature selection overview and categories

Unsupervised feature selection background

A Novel Unsupervised Feature Selection Approach Using Genetic Algorithm on Partitioned Data Amit Saxena Department of Computer Science and IT, Guru Ghashidash University, Bilashpur, India. Deepesh Chugh Department of Computer Engineering, Netaji Subhas Institute of Technology, Delhi, India. Himanshu Mittal [himanshu.mittal224@gmail.com](himanshu.mittal224@gmail.com) Department of AIDS, Indira Gandhi Delhi Technical University for Women, Delhi, India India Mohammad Sajid Department of Computer Science, Aligarh Muslim University, Aligarh, India. Ritu Chauhan Centre for computational Biology and Bioinformatics, Amity University, Noida, India. Eiad Yafi Centre for computational Biology and Bioinformatics, Amity University, Noida, India. Jian Cao Department of Computer Science and Engineering, Shanghai Jiaotong University, Shang-hai, China. Mukesh Prasad [Mukesh.Prasad@uts.edu.au](Mukesh.Prasad@uts.edu.au) School of Computer Science, FEIT, University of Technology Sydney, Sydney, Australia. Corresponding Author: Mukesh Prasad Copyright © 2022 Amit Saxena, et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 500 Citation: Amit Saxena, et al. A Novel Unsupervised Feature Selection Approach Using Genetic Algorithm on Partitioned Data. Advancesin Artificial Intelligence and Machine Learning. 2022;2(4):34 . [https://www.oajaiml.com/ | November-2022](https://www.oajaiml.com/ | November-2022) Amit Saxena, et al. Abstract A novel feature selection approach is presented in this paper. Sammon’s Stress Function transforms the high dimension data to a lower dimension data set. A data set is divided into small partitions. The features are assigned randomly to these partitions. Using GA with Sammon Error as fitness value, a small, desired number of features are selected from every partition. The combination of the reduced subsets of the features from these partitions is again divided into small partitions. After a certain number of iterating the process, a desired small number of features is obtained. For experimental validation, the proposed method has been tested on 11 standard datasets with three classifiers namely, Decision Tree, MLP and KNN. The classification accuracies obtained by the proposed method is highest on most of the considered datasets against the results reported in literature. Moreover, the proposed method selects comparatively less number of features in comparison to considered methods. The optimistic results obtained from the proposed method justify its strength. Keywords: Feature Extraction, Unsupervised Feature Selection, Genetic Algorithm, Classification Techniques. 1. INTRODUCTION With the introduction of sophisticated electronic gadgets, there has been a tremendous growth in capturing vivid features of a data. This has resulted in exponential increase in data size. However, it is seen that some features are redundant and do not contribute to the inference of a learnable model. In fact, such features degrade the performance of a learning model. For example, microarray-datasets like, leukemia dataset, are large datasets which consist of thousands of features. However, a significant portion of these datasets is redundant only. Thus, the abundance of redundant and un-significant features leads to the problem of ‘curse of dimensionality’ which is one of the major problems of data mining [1] and pattern classification [2] . To mitigate this, it is necessary to identify features which are not only important but dominant over the entire dataset in representing the dataset and helps in achieving better learnable model. These features are termed as representative features of the dataset. In literature, there are number of feature selection approaches to retain representative features of a data. These me

FAQs

What main idea does the proposed method use for unsupervised feature selection?

It transforms high-dimensional data using Sammon’s Stress Function, partitions the dataset, and then applies a genetic algorithm to select representative features without label information.

How does the genetic algorithm decide which features to keep?

It uses Sammon error as the fitness value, selecting a small desired number of features from each partition and iterating the process to obtain the final reduced set.

How is the method evaluated and what are the results?

It is tested on 11 standard datasets using Decision Tree, MLP, and KNN. The method achieves the highest accuracy on most datasets and selects comparatively fewer features than methods reported in the literature.

Aug 2, 2026

Uploaded by Rowan