BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Stochastic Models for Comparison-based Search
DTSTART:20170911T100000
DTEND:20170911T120000
DTSTAMP:20260407T195037Z
UID:8cac10a6e8e628a6ae4d70993f62727309b0905008940c51ea2d690e
CATEGORIES:Conferences - Seminars
DESCRIPTION:Daniyar Chumbalov\nEDIC candidacy exam\nExam president: Prof. 
 Boi Faltings\nThesis advisor: Prof. Matthias Grossglauser\nCo-examiner: Pr
 of. Michael Kapralov\n\nAbstract\nGiven a set of objects H with hidden fea
 tures in Rd\nwe are interested in finding a target object ~ 2 H by queryin
 g\npairwise comparisons of the objects in H to ~ and observing\ntheir nois
 y outcomes. Another problem we would like to explore\nis the actual embedd
 ing of the objects in H given the outcomes of\ntheir pairwise comparisons.
  Finally\, we are interested in solutions\nfor the combination of these tw
 o tasks into one reinforcement\nlearning problem.\nIn this proposal\, we d
 iscuss three papers related to our\nresearch. First\, we overview very gen
 eral bayesian active learning\nstrategies. Next we examine some ranking te
 chniques of preembedded\nobjects using pairwise comparisons. Finally\, we 
 discuss\nmodern ordinal embedding solutions.\n\nBackground papers\nNear-Op
 timal Bayesian Active Learning with Noisy Observations\, by Daniel Golovi
 n\, Andreas Krause\, Debajyoti Ray.\nActive Ranking using Pairwise Compari
 sons\, by Kevin G. Jamieson\, Robert D. Nowak.\nFinite Sample Prediction 
 and Recovery Bounds for Ordinal Embedding\, by Lalit Jain\, Kevin Jamieso
 n\, Robert Nowak
LOCATION:BC 02 https://plan.epfl.ch/?room==BC%2002
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR