Chaotic sequence and opposition learning guided approach for data clustering

Autor: Tribhuvan Singh, Nitin Saxena
Rok vydání: 2021
Předmět:
Zdroj: Pattern Analysis and Applications. 24:1303-1317
ISSN: 1433-755X
1433-7541
DOI: 10.1007/s10044-021-00964-2
Popis: Data clustering is a prevalent problem that belongs to the data mining domain. It aims to partition the given data objects into some specified number of clusters based on the sum of the intra-cluster distances. It is an NP-hard problem, and many heuristic approaches have already been proposed to target the desired objective. However, during the search process, the problem of local entrapment is prevalent due to nonlinear objective functions and a large range of search domains. In this paper, an opposition learning and chaotic sequence guided approaches are incorporated in a fast converging evolutionary algorithm called improved environmental adaptation method with real parameter (IEAM-R) for solving the data clustering problem. A chaotic sequence generated by a sinusoidal chaotic map has been utilized to target promising solutions in the search domain. On the other hand, the inclusion of the opposition learning-based approach allows the solutions to explore more appropriate locations in the search domain. The performance of the proposed approach is compared against some well-known algorithms using fitness values, statistical values, convergence curves, and box plots. These comparisons justify the efficacy of the suggested approach.
Databáze: OpenAIRE
Externí odkaz: