BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4//
BEGIN:VEVENT
UID:20261120T025216EST-9900ol8LMM@132.216.98.100
DTSTAMP:20261120T075216Z
DESCRIPTION:Title:\n\nSingular Learning Theory for PPCA and Factor Analysis
 \n\n \n\nAbstract:\n\nAn age-old question in principal components analysis
  (PCA) is how many principal directions should be retained? Model-based ap
 proaches\, like probabilistic PCA\, provide one way of addressing this que
 stion. Information criteria can be used to treat the choice of the number 
 of PCs as a model selection problem. However\, there is a subtlety that is
  often overlooked: the PPCA model\, as well as the related factor analysis
  model\, both have singularities\, which alter convergence rates that moti
 vate information criteria like the BIC. In this talk\, I will provide an o
 verview of singular learning theory\, an area that involves using ideas fr
 om algebraic geometry to tame the asymptotics of models with singularities
 . I will also introduce the singular BIC (sBIC) as an information criteria
  that is able to incorporate asymptotic information at singularities and c
 an be used for model selection. The efficacy of the sBIC as compared to th
 e BIC is demonstrated empirically and theoretically\, where it is shown th
 at in high-dimensional settings the sBIC requires a smaller eigengap in or
 der to detect principal component directions.\n\nThis project is joint wor
 k with Mathias Drton and Daniel Windisch.\n\nSpeaker\n\nAndy McCormack is 
 an assistant professor at the University of Alberta. He received his PhD f
 rom Duke University in 2023 under the supervision of Peter Hoff and subseq
 uently did a postdoc under Mathias Drton at the Technical University of Mu
 nich. He is interested in any problem where algebra\, geometry\, and stati
 stics come together to provide practical insights. In particular\, he is i
 nterested in the areas of non-Euclidean data analysis\, algebraic statisti
 cs\, information geometry\, and the analysis of tensor-valued data.\n\nonl
 ine via Zoom\n\nJoin Meeting\n\nhttps://mcgill.zoom.us/j/83483745684\n\n 
 \n\nMeeting ID: 834 8374 5684\n
DTSTART:20261120T203000Z
DTEND:20261120T213000Z
SUMMARY:Andy McCormack (University of Alberta)
URL:https://www.mcgill.ca/mathstat/channels/event/andy-mccormack-university
 -alberta-375660
END:VEVENT
END:VCALENDAR
