PRICAI 2016 Paper Selection | Differential Evolution Algorithm for Feature Selection and Case Selection

Feature Selection and Case Selection Differential Evolution Algorithm (A Differential Evolution Approach to Feature Selection and Instance Selection)

Abstract: Due to the continuous development of storage hardware and data acquisition technologies, more and more data are being collected. The incoming traffic of data is so huge that data mining technology cannot keep up with it. The collected data tends to have redundant or unrelated features/instances that limit the performance of the classification. Feature selection and instance selection help eliminate this problem by eliminating useless data. This paper proposes a series of algorithms using Differential Evolution to implement feature selection, instance selection, and the combination of feature selection and instance selection. The reduction of data, classification accuracy and training time are in line with the original data and existing algorithms. Experimental studies on ten different difficulty datasets have shown that newly developed algorithms can successfully reduce the size of data, and in most cases can maintain cooperation to increase classification performance. In addition, the calculation time also has a substantial reduction. This work is the first systematic, and a series of feature/instance selection algorithms are studied in the classification. The results show that the problem of instance selection is more difficult to solve than the problem of feature selection, but if the method is effective, the scale of data can be greatly reduced. And provide great benefits.

Keywords: differential evolution algorithm , feature selection , instance selection , classification

First author introduction

Bing Xue

Education: Doctor of Engineering and Computer Science, Victoria University of Wellington

Research directions: Artificial Intelligence, Machine Learning, Big Data/Connection Biology, Statistics, Engineering, and Mathematical Databases

Related academic papers:

· "A Survey on Evolutionary Computation Approaches to Feature Selection" (IEEE Transaction on Evolutionary Computation. Aug. 2016)

· "Cross-Domain Reuse of Extracted Knowledge in Genetic Programming for Image Classification" (IEEE Transaction on Evolutionary Computation 2016)

Via:PRICAI 2016

PS : This article was compiled by Lei Feng Network (search "Lei Feng Network" public number) and it was compiled without permission.

Original paper download

RandM Disposable Vape Pod Device

Hongkong Onice Limited , https://www.ousibangvape.com