­­­APLICACIÓN DE LA TEORÍA DE GRAFOS A LA OPTIMIZACIÓN DE REDES NEURONALES ARTIFICIALES

Abstract

The objective of the study is to explore how graph theory can be applied to optimize artificial neural networks, analyzing both the theoretical bases and the practical implications of this integration. In the theoretical framework, the author presents the foundations of artificial neural networks and graph theory, establishing their potential to model the connectivity of neural networks. The methodology combines literature review, theoretical analysis and practical experimentation optimizing neural networks with graph techniques. The results show significant improvements after optimization, reducing training time by 15% and increasing accuracy in tests by 10%. Furthermore, a correlation was found between network complexity and optimization benefits. The author concludes that the integration of graphs in the optimization of artificial neural networks improves efficiency and precision, with implications in applications such as natural language processing and image analysis. More research is required on variations in graph structure and size for different types of neural networks.