BEGIN:VCALENDAR
VERSION:2.0
PRODID:researchseminars.org
CALSCALE:GREGORIAN
X-WR-CALNAME:researchseminars.org
BEGIN:VEVENT
SUMMARY:Bryon ARAGAM (University of Chicago)
DTSTART:20250115T160000Z
DTEND:20250115T171500Z
DTSTAMP:20260423T021450Z
UID:AAIT/41
DESCRIPTION:Title: <a href="https://researchseminars.org/talk/AAIT/41/">Le
 arning compositional structure from data</a>\nby Bryon ARAGAM (University 
 of Chicago) as part of Seminar on Algorithmic Aspects of Information Theor
 y\n\n\nAbstract\nWe introduce the neighbourhood lattice decomposition of a
  distribution\, which is a compact\, non-graphical representation of condi
 tional independence that is valid in the absence of a faithful graphical r
 epresentation. The idea is to view the set of neighbourhoods of a variable
  as a subset lattice\, and partition this lattice into convex sublattices\
 , each of which directly encodes a collection of conditional independence 
 relations. We show that this decomposition exists in any compositional gra
 phoid and can be computed consistently in high-dimensions without the curs
 e of dimensionality. In particular\, this gives a way to learn from data a
 ll of the independence relations implied by any graphical model and thus i
 n particular its structure.\n
LOCATION:https://researchseminars.org/talk/AAIT/41/
END:VEVENT
END:VCALENDAR