BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Tackling Label Corruptions: Univariate Polynomial Regression and G
 eneralized Linear Models - Sushrut Karmalkar\, Microsoft Research
DTSTART:20260203T140000Z
DTEND:20260203T150000Z
UID:TALK243946@talks.cam.ac.uk
CONTACT:Fernando Ruiz Mazo
DESCRIPTION:Label corruptions pose a significant challenge in various mach
 ine learning tasks\, affecting the accuracy and reliability of models. In 
 this talk\, we will address two distinct problems involving label corrupti
 ons\, and present approaches to handle them effectively.\n\nThe first prob
 lem we consider is that of robust univariate polynomial regression. In thi
 s problem the goal is to recover a polynomial which is pointwise close to 
 a target polynomial\, given samples where\, with probability $\\alpha$ the
  samples are clean (satisfy the model)\; and with probability $1-\\alpha$ 
 the label is corrupted (completely arbitrary). We propose an approach whic
 h can tolerate a corruption fraction as large as any constant less than 1/
 2\, which is the information theoretic limit for unique recovery in this p
 roblem.\n\nIn the second problem\, we examine the challenge of learning a 
 linear function composed with a generalized linear model (GLM). We focus o
 n the oblivious noise setting\, where up to any constant fraction of the l
 abels are corrupted via arbitrary independent and additive noise. We show 
 that in this setting\, it is always possible to recover a polynomial-sized
  list of candidates\, one of which is arbitrarily close to the true answer
 . Furthermore\, under mild distributional assumptions\, we show this recov
 ery is unique. \n\n*This talk is co-hosted by the Computer Laboratory AI R
 esearch Group.*
LOCATION:FW 26\, Computer Laboratory\, William Gates Building
END:VEVENT
END:VCALENDAR