Vyhodnocení vlivu metod lokálního prohledávání v evolučních algoritmech

Abstract

The aim of the thesis is to analyse the impact of local search routines on evolutionary algorithms. Three discrete algorithms of Artificial Bee Colony, Differential Evolution and Harmony Search were analysed with embedded local search routines and tested on the flow shop with blocking problem. The tests were conducted on the benchmark data sets of Taillard for the first 90 instances. From the t-test analysis on the algorithms, it can be concluded that for all the algorithms, local search improves the generic algorithms significantly.