CSE DSI Machine Learning Seminar with Grigorios Chrysos (ECE, UW Madison)

Grounding Diffusion Models: The Path to Reliable Synthesis Without Hallucination

Diffusion models consistently produce stunningly realistic images, but they also frequently synthesize images that hallucinate. Building a reliable pipeline free of structural hallucinations forces us to ask key questions: can we prove that diffusion models will never hallucinate? Should we rely on deterministic or stochastic sampling for reliable generation? In the second part of the talk, we flip the question: are hallucinations always detrimental?

Grigorios Chrysos is an Assistant Professor at the University of Wisconsin-Madison, following the completion of his PhD at Imperial College London. Previously, he graduated from National Technical University of Athens with a Diploma/MEng in Electrical and Computer Engineering. Grigorios was awarded a rising star award by CPAL (2025) and a new faculty highlight award by AAAI (2026). His research interests lie in representation learning, tensor decompositions and designing trustworthy models. He has co-organized multiple workshops and tutorials on top-tier conferences, while he serves as Program Chair in CPAL’27.