20.500.12556/RUL-173262 Optimizacija parametrov in arhitekture nevronskih mrež s pomočjo genetskih algoritmov Optimization of Neural Network Parameters and Architectures Using Genetic Algorithms V diplomskem delu je predstavljena optimizacija arhitektur nevronskih mrež z uporabo genetskih algoritmov. Glavni izziv predstavlja sočasno določanje arhitekture in hiperparametrov, kar je pri klasičnih pristopih pogosto zamudno in neučinkovito. Razvit je bil kromosomski zapis, ki v enotni obliki združuje hiperparametre in parametre skritih plasti, ter postopki selekcije, mutacije in več načinov križanja. Definirana je bila funkcija uspešnosti, ki poleg uspešnosti učenja upošteva kazen za kompleksnost, ter progresivni urnik učenja. V nalogi smo oblikovali celovit algoritem za samodejno iskanje ustreznih arhitektur in hiperparametrov ter ga primerjali z algoritmom NEAT. Predlagani pristop smo ovrednotili na šestih podatkovnih zbirkah iz repozitorija UCI in pokazali, da dosega konkurenčne rezultate glede napovedne uspešnosti, vendar na račun večje kompleksnosti in časovne zahtevnosti v primerjavi z algoritmom NEAT, ki gradi enostavnejše arhitekture in je hitrejši. This thesis presents the optimization of neural network architectures using genetic algorithms. A key challenge in this area is the simultaneous determination of both network architecture and training hyperparameters, which in traditional approaches is often time-consuming and inefficient. To address this, we developed a chromosome representation that encodes hyperparameters and hidden layer parameters in a unified way, together with procedures for selection, mutation, and multiple crossover strategies. The approach incorporates a fitness function that balances predictive performance with a penalty for model complexity, as well as a progressive learning schedule. Based on these components, we designed a comprehensive algorithm for the automated search of suitable architectures and hyperparameters and compared it against the NEAT algorithm. The proposed method was evaluated on six benchmark datasets from the UCI repository, demonstrating competitive predictive performance, though at the cost of higher complexity and computational demand compared to NEAT, which produced simpler and faster models. genetski algoritmi nevronske mreže optimizacija arhitektur optimizacija hiperparametrov algoritem NEAT genetic algorithms neural networks architecture optimization hyperparameter optimization NEAT algorithm true false false Slovenski jezik Angleški jezik Diplomsko delo/naloga 2025-09-15 11:05:01 2025-09-15 11:05:08 2025-11-07 03:50:48 0000-00-00 00:00:00 2025 0 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 38043 250506243 Naglic_Vita_-_Optimizacija_parametrov_in_arhitekture_nevronskih_mrez_s_pomocjo_genetskih_algorit.pdf Naglic_Vita_-_Optimizacija_parametrov_in_arhitekture_nevronskih_mrez_s_pomocjo_genetskih_algorit.pdf 1 A812C1E58424F64E3ED6DD0070616705 1729800b8f9215054e852e3198d3d7a97b62e30bda451b427536be0d5d756703 e92d132e-9212-11f0-9328-0050569b8976 https://repozitorij.uni-lj.si/Dokument.php?lang=slv&id=217478 Fakulteta za računalništvo in informatiko 0 0 0