20.500.12556/RUL-120066 Primerjava in redukcija bio navdihnjenih metahevristik za optimizacijo Comparison and reduction of nature inspired metaheuristics for optimization Bio navdihnjena metahevristična optimizacija je zelo aktivno področje raziskav. V nalogi je narejen pregled bio navdihnjenih metahevrističnih algoritmov in razdelitev na podskupine. Izbral sem dva podobna algoritma (Optimizator sivih volkov in Algoritem optimizacije s kiti) in ju podrobno analiziral in primerjal. V primerjavi sem izpostavil ključne podobnosti in razlike obeh pristopov, ki sem jih nato prenesel v prevedbo. Implementiral sem eno prevedbo za vsak algoritem. Prevedbi sta vsebovali mehanizme za optimizacijo iz drugega algoritma. Testiranje je bilo izvedeno na dobro poznanih testnih funkcijah za optimizacijo. Pri testiranju sem opazil splošno poslabšanje učinkovitosti pri hibridih. Ugotovil sem, da različni mehanizmi za optimizacijo delujejo različno dobro na različnih testnih funkcijah in se večino časa ne mešajo najbolje. Vsebnost drugega algoritma lahko povzroči poslabšanje učinkovitosti, lahko povzroči izboljšanje, kjer bi se osnovni algoritem slabo odrezal, lahko pa tudi povroči močno poslabšanje, kjer bi osnovni algoritem dobil dober rezultat. Nature inspired metaheuristic optimization is a very active field of research. In this thesis I conducted an overview of nature inspired metaheuristics and made a comparison based on their features. I chose two very similar algorithms (Grey wolf optimizer and Whale optimization algorithm), made a detailed analysis of each one and a detailed comparison between the two. In the comparison I highlighted the key similarities and differences of both approaches which I then carried into the hybridization of the two. I implemented one hybrid for each of the algorithms. The hybrids contained mechanisms for optimizations from the other algorithm. Testing was done on well known test functions for optimization. In the results I noticed a general degradation of performance for the hybrid algorithms. I concluded that the different optimization mechanisms work with varying efficiencies for the different test functions and most of the time do not mix well. The presence of a different optimization mechanism from another algorithm can degrade performance in some cases, may improve it in others, where the original performs poorly, or may cause a significant degradation where the original performs well. metahevristka optimizacija bio-navdihnjnena metaheuristic optimization nature-inspired true false false Slovenski jezik Angleški jezik Diplomsko delo/naloga 2020-09-15 12:40:18 2020-09-15 12:40:23 2024-02-07 11:32:16 0000-00-00 00:00:00 2020 0 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 25744 31183875 Bracko_Bjorn_-_Primerjava_in_redukcija_bio_navdihnjenih_metahevristik_za_optimizacijo.pdf Bracko_Bjorn_-_Primerjava_in_redukcija_bio_navdihnjenih_metahevristik_za_optimizacijo.pdf 1 F52B9AE51235BB82B81FD9108F19DCD3 4bbe4edd5f1f6b6802814df7a071fc98bddb50bedf88cd2213ee047a8adfc5cf fd53355e-a1b9-11eb-a523-00155dcfd717 https://repozitorij.uni-lj.si/Dokument.php?lang=slv&id=134821 Fakulteta za računalništvo in informatiko 0 0 0