BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Impact of limited and noisy data on trustworthy machine learning
DTSTART:20221213T110000
DTEND:20221213T120000
DTSTAMP:20260916T070246Z
UID:8b7af09d75954c32950696fe2473ac6018e168fbc9857941992ac4a1
CATEGORIES:Conferences - Seminars
DESCRIPTION:Amartya Sanyal - ETH AI Center\nMachine Learning (ML) algorith
 ms are known to suffer from various issues when it comes to their trustwor
 thiness including properties like adversarial robustness\, robustness to d
 istribution shift\, privacy\, and fairness. Responsibly deploying ML algor
 ithms in critical applications requires satisfying these notions of trustw
 orthiness. While there has been significant progress in developing new tru
 stworthy algorithms\, the role of inadequate data is often ignored. Widely
  available data in the real world are often noisy\, limited\, and long-ta
 iled and play a role in hindering these aspects of trustworthiness. In thi
 s talk\, we will look at characterising some of the fundamental limitation
 s on trustworthiness due to inadequate data. In the second part of the tal
 k\, we will look at overcoming some of these limitations with plausible re
 laxations and new algorithms. Finally\, we will conclude with potential fu
 ture directions in this space.\n\n \n\nAmartya Sanyal is a postdoctoral f
 ellow at the ETH AI Center\, where he works with Prof. Fanny Yang and Prof
 . Bernhard Schölkopf. He completed his Ph.D in the Department of Computer
  Science at the University of Oxford with Prof. Varun Kanade and Prof. Phi
 lip H.S. Torr.  He has published papers in ICML\, ICLR\, NeurIPS\, and UA
 I and has also received multiple Spotlights as well as an Oral in these co
 nferences in addition to publishing an open problem in COLT. His main rese
 arch interests are in various aspects of trustworthy machine learning incl
 uding aspects of adversarial robustness\, privacy\, fairness\, and general
 isation. He is especially interested in understanding the theoretical and 
 empirical trade-offs between these properties and how to avert such trade-
 offs with better relaxations and approximations. In addition\, he is also 
 interested in understanding whether these theoretical results translate to
  practice in real world tasks.
LOCATION:ELA 002 https://plan.epfl.ch/?room==ELA%20002
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR