BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:3D de novo generation of organic molecules
DTSTART:20240625T151500
DTEND:20240625T161500
DTSTAMP:20260501T101248Z
UID:786f1c4b4d0c1af8069d65cc25e77a534a5d1471cc8876773c700b6c
CATEGORIES:Conferences - Seminars
DESCRIPTION:Ian is a PhD Candidate in the joint Carnegie Mellon University
  - University of Pittsburgh Computational Biology PhD Program where he is 
 advised by David Koes. His research is focused on developing deep generati
 ve models for applications in structure-based design.\nDeep generative mod
 els that can directly sample molecular structures with desired properties 
 have the potential to accelerate chemical discovery by reducing or elimina
 ting the need to engage in resource-intensive screening-based based discov
 ery paradigms. Designing deep generative models to accurately sample the c
 omplex distribution of 3D molecular structures is a challenging\, open pro
 blem. In this seminar we’ll present our recent work on adapting flow mat
 ching\, a modern generative modeling framework with connections to normali
 zing flows and diffusion\, for the task of 3D de novo generation of organi
 c molecules. This includes in exploration on how to extend flow matching t
 o model distributions of categorical variables. Additionally\, we will dis
 cuss our work in the broader context of an emerging class of flow matching
  methods for sampling molecular structures.
LOCATION:https://epfl.zoom.us/j/68447908297?pwd=OU5JUGJUSUhZc0ZNYjQ2WENvYl
 NRdz09
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR