I’m an ML researcher interested in generative modelling (esp. diffusion/flow models) as well as probabilistic inference, with a particular emphasis on sampling methods. My recent work has focused on applying ideas from probabilistic inference for post-training continuous and discrete diffusion models. I am also interested in the application of these techniques towards problems in the sciences (physics and biology in particular).
I’m currently a PhD student at Mila under the supervision of Yoshua Bengio. Previously I completed my Master’s Thesis with Pascal Poupart, with research focused on Bayesian federated learning methods.
For a list of my work see my Google scholar page
Contact me
mohsin[dot]hasan[at]mila[dot]quebec