Genetic-Algorithm-Based Instance and Feature Selection

Abstract

This chapter discusses a genetic-algorithm-based approach for selecting a small number of instances from a given data set in a pattern classification problem. Our genetic algorithm also selects a small number of features. The selected instances and features are used as a reference set in a nearest neighbor classifier. Our goal is to improve the classification ability of our nearest neighbor classifier by searching for an appropriate reference set. We first describe the implementation of our genetic algorithm for the instance and feature selection. Next we discuss the definition of a fitness function in our genetic algorithm. Then we examine the classification ability of nearest neighbor classifiers designed by our approach through computer simulations on some data sets. We also examine the effect of the instance and feature selection on the learning of neural networks. It is shown that the instance and feature selection prevents the overfitting of neural networks.

A Multi-Objective Evolutionary Algorithm Fitness Function for Case-Base Maintenance

“…We test the suitability of the approach on di↵erent datasets and compare the performance achieved to that of existing CBM algorithms from the literature. Previous works are mainly focused on reducing either the number of redundant cases or noisy cases [1,7,8,[20][21][22], or aimed at selecting attributes [10,12] or to both enhance the accuracy and reduce the size of the case-base [11]. However, the fitness function proposed in this work measures the redundancy of the case-base, the number of noisy cases and the error rate of the system.…”

Adaptive data reduction for large-scale transaction data

“…An initial population of chromosomes are selected either randomly or with some heuristics, and the selection-crossover-mutation operations are applied to the chromosomes to obtain better solutions. Ishibuchi [13] attempted to apply the GA to both feature and instance selection simultaneously and did obtain some good results, but the study is limited to the problem domain where the nearest neighbor method is applied for classification analysis on numeric data. In short, it appears that existing studies on applying the GA for data reduction tend to be task-specific (classification in particular), and whether and how the GA techniques can be used for general purpose data reduction remains to be investigated.…”

“…The application of the GAs for feature selection has been explored in [5,13,23], among others. Since feature selection is often associated with classification problems, the fitness functions specified in these studies are more or less a function of classification error rate.…”

“…Many task-specific data reduction approaches use adaptive methods, rather than random sampling [5,[23][24][25]. In particular, a genetic algorithm based search procedure for data reduction is introduced in [13] for classification analysis. An overview of data reduction techniques can be found in [19].…”

Genetic algorithms for outlier detection and variable selection in linear regression models

“…Finally, a related area of research is that of the use of GAs for feature and instance selection in, for example, data mining. A number of references to this literature can be found in [10].…”