Adaptácia parametrov samoorganizujúceho sa migračného algoritmu pomocou neurónových sietí

Abstract

In recent years, there have been many improvements regarding the Self-Organizing Migrating Algorithm. The aim of my bachelor thesis was to find out if it is possible to improve this algorithm by adding a neural network to get a new algorithm that will work more efficiently by selecting individual parameter settings for each individual during the migration loop. The neural network is learned by using the Self-Organizing Migrating Algorithm. The new algorithm should decide on the values of individual parameters for the selected individual. It turned out that the neural network sets these values very similarly for the specific functions. From the resulting values, we found that the algorithm together with the neural network achieves similar to better results on individual test functions.