Epistemic Generative Adversarial Networks
This addresses the issue of GANs generating similar samples rather than diverse variations, which is important for applications requiring representative data generation, though it appears incremental as it builds on existing GAN frameworks.
The paper tackles the problem of limited output diversity in Generative Adversarial Networks (GANs) by introducing a novel loss function based on Dempster-Shafer theory and an architectural enhancement to quantify uncertainty, resulting in improved generation variability.
Generative models, particularly Generative Adversarial Networks (GANs), often suffer from a lack of output diversity, frequently generating similar samples rather than a wide range of variations. This paper introduces a novel generalization of the GAN loss function based on Dempster-Shafer theory of evidence, applied to both the generator and discriminator. Additionally, we propose an architectural enhancement to the generator that enables it to predict a mass function for each image pixel. This modification allows the model to quantify uncertainty in its outputs and leverage this uncertainty to produce more diverse and representative generations. Experimental evidence shows that our approach not only improves generation variability but also provides a principled framework for modeling and interpreting uncertainty in generative processes.