Using support vector machine and dynamic parameter encoding to enhance global optimization

Hybrid evolutionary algorithm with extreme machine learning fitness function evaluation for two-stage capacitated facility location problems

“…Besides incorporating successful elements of the previously constructed effective heuristic algorithms (Fernandes, Rocha, Aloise, Ribeiro, Santos & Silva, 2014), the proposed HEA gains pretty good balance between quality of solutions and computational time in two respects: 1) incorporating local search strategy to enhance the quality of the elitist individuals, and 2) introducing approximating fitness evaluation approach based on the ELM to alleviate the computational burden of the HEA. The idea of introducing fitness approximation has been proven to be effective and efficient in solving complex continuous optimization problems (Jin, Olhofer & Sendhoff, 2002;Hacioglu, 2007;Zheng, Chen, Liu & Huang, 2016).…”

“…Various techniques for the construction of surrogate models (also called approximation models or meta-models) have been employed to obtain the efficient and effective hybrid EAs. Among these techniques, artificial neural network (ANN), support vector machine (SVM) and kriging models are among some of the most prominent and commonly used techniques (Hacioglu, 2007;Zhang, Liu, Tsang & Virginas, 2010;Dias, Rocha, Ferreira & do Carmo Lopes, 2014;Zheng, Chen, Liu & Huang, 2016). By elaborating surrogate models, the computational burden can be greatly reduced.…”