BEGIN:VCALENDAR
PRODID:-//eluceo/ical//2.0/EN
VERSION:2.0
CALSCALE:GREGORIAN
BEGIN:VEVENT
UID:www.tcs.tifr.res.in/event/1427
DTSTAMP:20240521T095005Z
SUMMARY:Learning in Large and Structured Environments: Algorithms\, Guarant
 ees and Applications
DESCRIPTION:Speaker: Sayak Chowdhury (Microsoft Research Lab\, India)\n\nAb
 stract: \n\nIn this talk\, I will present some of my work on designing and
  analyzing algorithms for learning in large and structured environments\, 
 where the state and action spaces are huge or even infinite. I will focus 
 on two main topics: (1) Bayesian optimization for hyperparameter tuning in
  large-scale machine learning models\, and (2) Policy optimization for lan
 guage models using human feedback. For the first topic\, I will introduce
  the Gaussian process optimization framework and design multi-armed-bandi
 t algorithms for hyperparameter optimization. I will show sublinear regre
 t bounds for the proposed algorithms that depend on the information comple
 xity of the objective function to be optimized. Along the way\, I will pr
 esent a self-normalized concentration inequality for vector-valued marting
 ales of arbitrary\, possibly infinite\, dimension\, and discuss some appli
 cations of this concentration bound. For the second topic\, I will talk a
 bout the effects of noisy preference data that can negatively impact lang
 uage model alignment. I will propose a robust loss function for language 
 model policy optimization in the presence of random preference flips. I w
 ill show that the proposed language model policy is provably tolerant to 
 noise and characterize its sub-optimality gap as a function of noise rate\
 , dimension of the policy parameter\, and sample size. I will also demons
 trate the empirical performance of the proposed policy on various tasks\, 
 such as dialogue generation and sentiment analysis. I will conclude with 
 some open problems and future directions of research in large scale machin
 e learning.\n \nShort Bio:\nSayak Ray Chowdhury is a postdoctoral researc
 her at Microsoft Research\, India. Prior to this he was a postdoctoral fel
 low at Boston University\, USA. He obtained his PhD from the Dept of ECE\,
  Indian Institute of Science\, where he was a recipient of Google PhD fell
 owship. His research interests include reinforcement learning\, Bayesian o
 ptimization\, multi-armed bandits and differential privacy. Recently\, he 
 has been working towards mathematical and empirical understandings of lang
 uage models. More details about his research can be found here: https://si
 tes.google.com/view/sayakraychowdhury/home\n \n
URL:https://www.tcs.tifr.res.in/web/events/1427
DTSTART;TZID=Asia/Kolkata:20240529T160000
DTEND;TZID=Asia/Kolkata:20240529T170000
LOCATION:A-201 (STCS Seminar Room)
END:VEVENT
END:VCALENDAR