Grad Seminar 2026-2027
Queen's Graduate Mathematics & Statistics Society
Schedule:
-
(Friday, Oct 9, 1:00-2:00) Evan Arsenault, The role of Differential Equations in Generative AI.
Abstract: The success of many Generative AI advancements can be explained, at least in part, by the development of tractable training objectives for new models. This talk focuses on one such technique, flow-matching, which has found success in image generation tasks. In this talk, I will present the concept of flow-matching and highlight the clever probabilistic tools that make it so effective. Time permitting, I will demonstrate some simple computational examples.
-
(Friday, Sep 25, 1:00-2:00) Trevor Shillington, Jacobian and Dixmier Conjectures.
Abstract: The recent influx of AI-done proofs in math, in some sense, was kick-started with the disproof of the Jacobian conjecture this summer. More interestingly, at least to me, is the unintended disproof of the Dixmier conjecture. In this talk, I will develop the language needed to understand both the Jacobian and Dixmier conjectures, as well as shine a light on how the Dixmier conjecture was disproved.
If you want to give a talk, suggest a topic, or have any question, please contact Samuel Leblanc or Trevor Shillington.