ProbML 2026

Symposium on Probabilistic Machine Learning

(Previously Symposium on Advances in Approximate Bayesian Inference)

Co-located with ICML 2026 in South Korea

July 5, 2026

Venue: Korea Institute for Advanced Study Auditorium

Photo by Larry Koester, CC BY 2.0, via Flickr

Attendance Information

Registration

ProbML 2026 is a free event, but spots are limited! Make sure you register as soon as possible.

Register for ProbML 2026

We have limited space, please cancel your registration if you are unable to attend! This will allow others to register.

Venue

ProbML 2026 will be hosted in the KAIST Auditorium at the Korea Institute for Advanced Study (KAIST) in Seoul.

KAIST College of Business
85 Hoegi-ro, Dongdaemun District
Seoul, South Korea

Schedule

TimeEventNote
08:00 – 09:00Check-In
09:00 – 09:12Opening Remarks
09:12 – 09:15Sponsor Remarks

Session 1 — Chair: TBD
09:15 – 10:00Invited Talk 1A Bayesian Nonparametric Approach to Adapting Model Size During Training
Guiomar Pescador-Barrios
10:00 – 10:15Contributed Talk 1An Isotropic Approach to Efficient Uncertainty Quantification with Gradient Norms
Nils Grünefeld, Jes Frellsen, Christian Hardmeier
10:15 – 10:45Coffee Break

Session 2 — Chair: TBD
10:45 – 11:30Invited Talk 2From Drift to Coherence: Stabilizing Beliefs in LLMs
Juho Lee
11:30 – 11:45Contributed Talk 2CogFormer: Learn All Your Models Once
Jerry M. Huang, Lukas Schumacher, Niek Stevenson, Stefan T. Radev
11:45 – 14:00Lunch

Session 3 — Chair: TBD
14:00 – 14:45Invited Talk 3TBA
TBA
14:45 – 16:45Poster SessionRefer to the accepted papers list for poster numbers.
16:30 – 16:45Coffee Break
16:45 – 16:55Invited Position Talk 1TBA
TBA
16:55 – 17:05Invited Position Talk 2TBA
TBA
17:05 – 17:45Panel DiscussionPanelists: TBA
17:45 – 18:00Closing Remarks
18:00 – open endTown Hall Discussion