BEGIN:VCALENDAR
PRODID:-//eluceo/ical//2.0/EN
VERSION:2.0
CALSCALE:GREGORIAN
BEGIN:VEVENT
UID:www.tcs.tifr.res.in/event/908
DTSTAMP:20230914T125943Z
SUMMARY:Conditional Independence : Novel Consistent Estimators & Hypothesis
  Testing Paradigm
DESCRIPTION:Speaker: Himanshu Asnani (Department of Electrical Engineering\
 nUniversity of Washington\nSeattle\, Washington.)\n\nAbstract: \nAbstract
 : The problem of ascertaining conditional independence or dependence is ce
 ntral to causal discovery and statistical inference in several dynamical s
 ystems\, such as gene regulatory networks\, finance networks\, edge testin
 g in Bayesian Networks etc. In this talk\, we navigate this problem via tw
 o statistical approaches.\nIn the first approach\, for the low dimensional
  regime\, we develop consistent sample estimators  based on nearest neigh
 bor methods for conditional mutual information (CMI) in general probabilit
 y spaces\, that is\, even when the variables are mixtures of continuous an
 d discrete components or have only low-dimensional manifolds. In general\,
  we define a general graph divergence measure (GDM)\, as a measure of inco
 mpatibility between the observed distribution and a given graphical model 
 structure. This generalizes to estimating several multivariate information
  measures\, a special case of which is conditional mutual information. We 
 construct a novel estimator via a coupling trick that directly estimates t
 hese multivariate information measures using the Radon-Nikodym derivative.
 \n\nFor the second approach\, in the case of high dimensional regime\, we 
 study the conditional independence hypothesis testing problem\, where we d
 evelop a new "mimic and classify" paradigm which is realized in two-steps:
  (a) mimic the conditionally independent (CI) distribution close enough to
  recover the support\, and (b) classify to distinguish the joint and the C
 I distribution. Thus\, as long as we have a good generative model and a go
 od classifier\, we potentially have a sound CI Tester. With this modular p
 aradigm which has provable p-value guarantees\, CI Testing also becomes am
 iable to be handled by state-of-the-art\, both generative and classificati
 on methods from the modern advances in Deep Learning.\nBoth the above appr
 oaches are benchmarked on synthetic and real datasets. Finally\, with less
 ons drawn from these two approaches\, we present future research program\,
  broadly\, at the intersection of Information Theory and Statistical Learn
 ing (including modern Deep Learning methods) and their applications.\nBio:
  Dr. Himanshu Asnani is currently a Research Associate in the Electrical 
 Engineering Department at University of Washington\, Seattle and Visiting 
 Assistant Professor in the Electrical Engineering Department at IIT Bombay
 . His research interests include information and coding theory\, statistic
 al learning and inference and machine learning. He has been named Amazon C
 atalyst Fellow for the year 2018. Dr. Asnani is the recipient of 2014 Mar
 coni Society Paul Baran Young Scholar Award. He received his Ph.D. in Elec
 trical Engineering Department in 2014 from Stanford University\, working u
 nder Professor Tsachy Weissman\, where he was a Stanford Graduate Fellow. 
 Following his graduate studies\, he worked in Ericsson Silicon Valley as a
  System Architect for couple of years\, focusing on designing next generat
 ion networks with emphasis on network redundancy elimination and load bala
 ncing. Driven by a deep desire to innovate and contribute in the education
  space\, with the aid of technology\, Dr. Himanshu Asnani quit his corpora
 te sojourn and got involved for a while in his education startups (where h
 e currently holds Founding Advisor role) to bring the promise of quality e
 ducation in vernacular languages in underdeveloped and developing countrie
 s - places which do not have access to English\, Internet and Electricity.
  In the past\, he has also held visiting faculty appointments in the Elect
 rical Engineering Department\, Stanford University. He was the recipient o
 f Best Paper Award at MobiHoc 2009 and was also the finalist for Student P
 aper Award in ISIT 2011\, Saint Petersburg\, Russia. Prior to that\, he re
 ceived his B.Tech. from IIT Bombay in 2009 and M.S. from Stanford Universi
 ty in 2011\, both in Electrical Engineering.\n
URL:https://www.tcs.tifr.res.in/web/events/908
DTSTART;TZID=Asia/Kolkata:20181015T143000
DTEND;TZID=Asia/Kolkata:20181015T153000
LOCATION:A-201 (STCS Seminar Room)
END:VEVENT
END:VCALENDAR