Application of generalized f-divergences in deep learning

This bachelor thesis will focus on the practical application of generalized f-divergences based on Bayes spaces as training objectives in artificial intelligence, specifically in generative adversarial networks (GANs). An important part of the work will be a comparison with divergences commonly used in GAN training, including an analysis of the potential advantages and limitations of the individual approaches. For this purpose, real-world data from publicly available testing and benchmark datasets will be used.