Gen AI Hub at ICML

The ICML conference takes place in Vancouver, Canada (Mike Benna@mbenna)

Hub researchers are well represented at this year’s International Conference on Machine Learning (ICML) in Vancouver.

Three papers on diffusion models co-authored by Methodologies working group co-lead Arthur Gretton (UCL/Deep Mind) have been accepted for poster sessions:

• Distributional Diffusion Models with Scoring Rules 15 July, -- Poster Session 1

• Accelerated Diffusion Models via Speculative Sampling 16 July -- Poster Session 2

• Distributional Diffusion Models with Scoring Rules 16 July -- Poster Session 3

Jason Hartford (University of Manchester) from the Gen Biology working group is involved with two papers presented and is also giving a keynote at the Scaling Up Intervention Models workshop on 18 July.

The papers will appear in the following poster sessions:

• ViTally Consistent: Scaling Biological Representation Learning for Cell Microscopy - 15 July

• Towards scientific discovery with dictionary learning: Extracting biological concepts from microscopy foundation models - 17 July

Pasquale Minervini (University of Edinburgh), a lead of the Multi Modal Models working group, has two papers in the main conference:

• Is Complex Query Answering Really Complex?

• When Can Proxies Improve the Sample Complexity of Preference Learning?

Jagmohan Chauhan (UCL), a member of several hub working groups, is presenting two papers at the conference, including Continual Generalized Category Discovery: Learning and Forgetting from a Bayesian Perspective.

Hub director David Barber is a co organiser of a full day workshop on computer use agents on 19 July. You can find the full details on this web page.

If you can’t make it to Vancouver, you can register for a virtual pass allowing you to view the ICLM live stream using this link.