Schedule

November 8-10 2023

Register

The times are given in the local Montreal time (Eastern Standard Time, UTC-5).

The conference will take place at the Agora of Mila, which is located at 6650, Rue Saint-Urbain, Montreal, QC, H2S 3H1.

Day 1 - Background - November 8th 2023

This session is designed to provide pre-requisites on graphs and probability theory. It aims to level the audience who may feel a little behind on these topics or would like a refresher. It covers the essential requirements to understand the material presented on Day 2 and Day 3.

Time (EST) Session Speaker Title
09:00 - 09:30 Registration
09:30 - 09:45 Opening Remarks
09:45 - 10:30 Talk Vitória Barin Pacela Introduction to Probability
10:30 - 11:00 Break
11:00 - 11:45 Talk Tristan Deleu Probabilistic Modeling
11:45 - 12:30 Talk Sébastien Lachapelle Probabilistic Inference
12:30 - 14:00 Lunch
14:00 - 14:45 Talk Pablo Lemos Sampling Methods
14:45 - 15:30 Talk Padideh Nouri Sampling with Reinforcement Learning

Day 2 - Fundamentals - November 9th 2023

The focus is to provide motivation and introduce basic to moderate theory related to GFlowNets. The sessions will delve into the fundamental concepts and principles, ensuring a solid theoretical foundation.

Both practical sessions can be done on Google Colab ( Session 1 Problem Set, Session 1 Solutions, Session 2 Problem Set, Session 2 Solutions, ), or locally using the IPython notebooks found on GitHub,

Time (EST) Session Speaker Title
08:30 - 09:00 Registration
09:00 - 09:30 Talk Yoshua Bengio Opening Remarks
09:30 - 10:15 Talk Emmanuel Bengio Introduction to GFlowNets: Part 1
10:15 - 10:45 Talk Nikolay Malkin Introduction to GFlowNets: Part 2
10:45 - 11:00 Break
11:00 - 11:30 Talk Jarrid Rector-Brooks Training GFlowNets
11:30 - 12:30 Talk Joseph Viviano Hands-on: Training GFlowNets
12:30 - 14:00 Lunch
14:00 - 14:40 Talk Moksh Jain Parameterization and Conditioning
14:40 - 15:00 Talk Léna Néhale Ezzine Maximum Likelihood Training
15:00 - 15:30 Break
15:30 - 16:15 Talk Salem Lahlou Continuous GFlowNets
16:15 - 16:30 Break
16:30 - 17:30 Talk Joseph Viviano Hands-on: Continuous GFlowNets
17:30 - 20:30 Social

Day 3 - Applications - November 10th 2023

This section will include a code tutorial, practical applications, and engaging discussions. Participants will have the opportunity to apply the concepts learned on Day-1 to real-world scenarios, enhancing their understanding through hands-on practice.

Time (EST) Session Speaker Title
08:30 - 09:00 Registration
09:00 - 11:15 Short talks GFlowNets for scientific discovery
09:00 - 09:15 Talk Introduction
09:15 - 09:30 Talk Moksh Jain Biological sequence design
09:30 - 09:45 Talk Emmanuel Bengio Molecular design
09:45 - 10:00 Talk Léna Néhale Ezzine, Alexandra Volokhova Molecular and protein conformation generation
10:00 - 10:15 Break
10:15 - 10:30 Talk Alex Hernandez-Garcia Materials discovery
10:30 - 10:45 Talk Mizu Nishikawa-Toomey Causal discovery
10:45 - 11:00 Talk Dinghuai Zhang Combinatorial optimisation
11:00 - 11:15 Talk Nikolay Malkin Language models
11:15 - 11:30 Break
11:30 - 12:30 Talk Emmanuel Bengio Evaluation metrics for GFlowNets
12:30 - 14:00 Lunch
14:00 - 15:00 Tutorial Alex Hernandez-Garcia Live coding of a GFlowNet environment
15:00 - 15:15 Break
15:15 - 16:30 Interactive Organisers GFlowNets Clinic: what it is for and what it is not
16:30 - 16:45 Break
16:45 - 17:30 Talk Yoshua Bengio Open problems and future perspective