BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Identifiable Exchangeable Mechanisms for Causal Structure and Repr
 esentation Learning - Dr. Miles Cranmer (DAMTP/IoA Assistant Prof.)  
DTSTART:20251125T153000Z
DTEND:20251125T163000Z
UID:TALK240913@talks.cam.ac.uk
CONTACT:Liz Tan
DESCRIPTION:*Identifiable Exchangeable Mechanisms for Causal Structure and
  Representation Learning (Dr. Miles Cranmer)*: Identifying latent represen
 tations or causal structures is important for good generalization and down
 stream task performance. However\, both fields have been developed rather 
 independently. We observe that several methods in both representation and 
 causal structure learning rely on the same data-generating process (DGP)\,
  namely\, exchangeable but not i.i.d. (independent and identically distrib
 uted) data. We provide a unified framework\, termed Identifiable Exchangea
 ble Mechanisms (IEM)\, for representation and structure learning under the
  lens of exchangeability. IEM provides new insights that let us relax the 
 necessary conditions for causal structure identification in exchangeable n
 on--i.i.d. data. We also demonstrate the existence of a duality condition 
 in identifiable representation learning\, leading to new identifiability r
 esults. We hope this work will pave the way for further research in causal
  representation learning. Discussing the paper P. Reizinger\, S. Guo\, F. 
 Huszar et al. (2024) https://arxiv.org/abs/2406.14302
LOCATION:MR4\, Centre for Mathematical Sciences\, Wilberforce Rd\, Cambrid
 ge CB3 0WA
END:VEVENT
END:VCALENDAR