BEGIN:VCALENDAR
PRODID:-//eluceo/ical//2.0/EN
VERSION:2.0
CALSCALE:GREGORIAN
BEGIN:VEVENT
UID:www.tcs.tifr.res.in/event/1541
DTSTAMP:20250408T091523Z
SUMMARY:Corruption-Tolerant Algorithms for Generalized Linear Model
DESCRIPTION:Speaker: Debojyoti Dey (IIT Kanpur)\n\nAbstract: \nGeneralized 
 Linear Model (GLM) is a unified framework which brings Linear regression o
 ver real valued labels\, Gamma or Poisson regression over positive labels\
 , as well as Logistic regression over Binary labels under the same umbrell
 a\, and solves a Maximum Likelihood Estimation (MLE) problem to estimate t
 he model generating the data. Even unsupervised learning model such as Emp
 irical Mean Estimator is an example of maximum likelihood estimate. The ML
 E based algorithms often fail to recover the true model when a fraction of
  the observed data points are adversarially contaminated. \nIn this talk\
 , I am going to discuss our work on Robust Learning Algorithms for GLM und
 er adversarial corruptions. In this work\, we introduced a version of Expe
 ctation Maximization (EM) algorithm which exploits an adaptive variance al
 teration while solving a weighted MLE. The algorithm\, called SVAM (Sequen
 tial Variance-Altered MLE )\, offers provable model recovery guarantees su
 perior to the state-of-the-art for robust regression even when a constant 
 fraction of training labels are corrupted. The algorithm is also efficient
  in the sense that it offers linear rate of convergence to true optima. Ap
 art from linear regression\, the technique and the result extend to gamma 
 and logistic regression\, mean estimation etc. SVAM also empirically outpe
 rforms several existing problem-specific techniques for robust regression 
 and classification.\nThe talk is based upon a published article\, coauthor
 ed by Bhaskar Mukhoty and Purushottam Kar.\nShort Bio:\nDebojyoti Dey is a
  final year PhD student at IIT Kanpur. His research interest spans non-con
 vex and robust optimization\, distribution learning in Probabilistic Graph
 ical Model etc.\n
URL:https://www.tcs.tifr.res.in/web/events/1541
DTSTART;TZID=Asia/Kolkata:20250408T160000
DTEND;TZID=Asia/Kolkata:20250408T170000
LOCATION:A-201 (STCS Seminar Room)
END:VEVENT
END:VCALENDAR