Implementace a vizualizace algoritmu EDA

Abstract

This bachelor’s thesis focuses on the design, implementation, and visualization of Estimation of Distribution Algorithms (EDA), which represent a specific approach to evolutionary optimization. EDA operate by modeling a probability distribution over a set of high-quality solutions. The goal of this work is to design a general framework for such algorithms that allows easy extension with additional variants, and to develop tools for visualizing their behavior during the optimization process. As part of the solution, a generic EDA framework was first developed, from which three specific implementations were derived: an algorithm based on Bayesian networks (BOA), an algorithm using Gaussian and Student’s copula functions, and a hybrid approach combining EMNA and CMA-ES. Each variant was implemented as a standalone module following a unified interface, ensuring clarity and future extensibility. The proposed system was tested on a set of standardized benchmark functions from the CEC 2014 suite, which serve to evaluate the ability of algorithms to solve complex optimization problems. The test results provide feedback on the effectiveness of the approach and the usefulness of the implementation. The visualization outputs contribute to a deeper understanding of the dynamics and progression of solutions during the optimization process.