The application of social-spider optimization for parameter improvement in the Lukasiewicz structure
Jintara Dollaor, Sirapat Chiewchanwattana, Khamron Sunat, Nipotepat Muangkote · 2016
Feature selection for the optimal parameters within the Lukasiewicz structure is critical, as varied parameter values may degrade its performance, as well as increase the computational burden involved. In this paper, we introduce Social-Spider Optimization (SSO) to determine suitable values of similarity, in conjunction with fuzzy-entropy evaluation, in order to select data features; and compare the results obtained with both Particle Swarm Optimization (PSO) and Quantum Particle Swarm Optimization (QPSO). The experimental results indicate improvement through SSO in general accuracy and performance, over comparative techniques for parameter values in the Lukasiewicz structure.