Generative Models and Uncertainty Quantification

Copenhagen, September 14 and 15, 2022

Generative models and uncertainty quantification lie at the heart of Bayesian modelling and inference. At this small meeting, we discuss recent developments within the field. The meeting is deliberately kept small in order to ensure that discussion remains honest, lively and interesting. Attendance is, thus, mostly by invitation, but one can apply to join (see below).

Speakers

Tim Salimans
Machine Learning research scientist, Google Brain.

Vincent Dutordoir
PhD Candidate, University of Cambridge.

Yingzhen Li
Lecturer, Imperial College London.

Andrew Gordon Wilson
Associate Professor, Courant Institute of Mathematical Sciences, New York University.

Brooks Paige
Associate professor, University College London.

James Hensman
Principal researcher, Microsoft Research Cambridge.

Carl Henrik Ek
Associate professor, University of Cambridge.

Philipp Hennig
Professor, University of Tübingen.

Emiel Hoogeboom
PhD student, University of Amsterdam.

Cheng Zhang
Principal researcher, Microsoft Research Cambridge.

Antoine Wehenkel
PhD student, University of Liège.

Benjamin Bloem-Reddy
Assistant professor, University of British Columbia.

Arno Solin
Assistant Professor, Aalto University.

Program

Wednesday (Sep 14)		Thursday (Sep 15)
Session 1 Learning with limited supervision Chair: Jes Frellsen		Session 5 Diffusion models Chair: Ole Winther
8:55-9:00	Opening remarks Jes Frellsen
9:00-10:00	Yingzhen Li Understanding masked pre-training: fundamentally different from "old-fashioned" unsupervised learning?	9:00-10:00	Emiel Hoogeboom Score-based diffusion: how did we get here and open problems
10:00-10:40	Brooks Paige Active learning with semi-supervised learners	10:00-10:40	Vincent Dutordoir Neural Diffusion Processes
10:40-11:00	Coffee	10:40-11:00	Coffee
Session 2 Identifiability Chair: Søren Hauberg
11:00-12:00	Benjamin Bloem-Reddy Why (and when) should we care about identifiability in generative models?	11:00-12:00	Tim Salimans Imagen: diffusion models for text2image and beyond
12:00-13:00	James Hensman Orthogonality and identifiability in probabilistic models	12:00-13:00	Arno Solin Generative modelling with inverse heat dissipation
13:00-14:00	Lunch	13:00-14:00	Lunch
Session 3 Statistical and causal inference Chair: Pierre-Alexandre Mattei		Session 6 Scientific applications of generative models Chair: Wouter Boomsma
14:00-15:00	Cheng Zhang Causal Decision Making under Uncertainty	14:00-15:00	Antoine Wehenkel The symbiosis between deep probabilistic and scientific models
15:00-16:00	Philipp Hennig Inference through simulations with Differential Equation Filters	15:00-16:00	Coffee and concluding remarks
16:00-16:20	Coffee
Session 4 Scientific applications of generative models Chair: Wouter Boomsma
16:20-17:20	Carl Henrik Ek Generative Model for Sequential Alignment
17:20-18:50	Poster Session
19:00-	Workshop Dinner at restaurant Carl's Øl og Spisehus (directions)

Registration

If you would like to attend this workshop but have yet to receive an invitation, then please write an email to Søren Hauberg and include a link to your website or Google scholar profile. Note that seating is limited, so we cannot guarantee a ticket.

Venue

The workshop takes place at:

Carlsberg Akademi
Gamle Carlsberg Vej 15
1799 Copenhagen V
Venue website

Venue

Organizers

The conference is jointly organized by:

This workshop is funded through the Center for Basic Machine Learning Research in Life Science (NNF20OC0062606), the Pioneer Centre for AI (P1), the European Research Council (757360), the Independent Research Fund Denmark (9131-00082B) and the Novo Nordisk Foundation (NNF20OC0065611).