BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Bandit Algorithms for Online Matrix Factorisation
DTSTART:20230830T100000
DTEND:20230830T120000
DTSTAMP:20260406T202307Z
UID:9a197e4a6df9615362eaebbf66d2060b40dd9f3dd016fd3120673824
CATEGORIES:Conferences - Seminars
DESCRIPTION:Oscar Villemaud\nEDIC candidacy exam\nExam president: Prof. Ni
 colas Flammarion\nThesis advisor: Prof. Matthias Grossglauser\nCo-examiner
 : Prof. Olivier Lévêque\n\nAbstract\nMatrix factorisation is widely used
  for recommender systems\, but it is typically used in a offline manner\, 
 which does not leverage the information we gain when an item is recommende
 d. On the other hand\, linear bandit algorithms update recommendations wit
 h every new data point\, but they usually assume the exact knowledge of on
 e of the two factors. In this work\, we study how to perform matrix factor
 isation in an online manner without knowing any of the factors.\n\nBackgro
 und papers\n- Ruslan Salakhutdinov and Andriy Mnih. Bayesian probabilist
 ic matrix factorization using markov chain monte carlo. ICML\, 2008\n- Ab
 basi-Yadkori\, Y.\, Pál\, D.\, & Szepesvári\, C.\, Improved algorithms 
 for linear stochastic bandits\, NeurIPS\, 2011\n- Jaya Kawale\, Hung H B
 ui\, Branislav Kveton\, Long Tran-Thanh\, and Sanjay Chawla. Efficient th
 ompson sampling for online matrix-factorization recommendation\, NIPS\, 20
 15
LOCATION:BC 010 https://plan.epfl.ch/?room==BC%20010
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR