BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Representation Learning: A Causal Perspective - Yixin Wang
DTSTART:20211104T153000Z
DTEND:20211104T170000Z
UID:TALK163777@talks.cam.ac.uk
CONTACT:Chaochao Lu
DESCRIPTION:Representation learning constructs low-dimensional representat
 ions to\nsummarize essential features of high-dimensional data like images
  and\ntexts. Ideally\, such a representation should capture non-spurious\n
 features of the data in an efficient way. It shall also be\ndisentangled s
 o that we can freely manipulate each of its dimensions.\nHowever\, these d
 esiderata are often intuitively defined and\nchallenging to quantify or en
 force.\n\nIn this talk\, we take on a causal perspective of representation
 \nlearning. We show how desiderata of representation learning can be\nform
 alized using counterfactual notions\, which then enables algorithms\nthat 
 target efficient\, non-spurious\, and disentangled representations\nof dat
 a. We discuss the theoretical underpinnings of the algorithm and\nillustra
 te its empirical performance in both supervised and\nunsupervised represen
 tation learning.\n\nSpeaker Bio:\n\nI am an LSA collegiate fellow and an a
 ssistant professor of Statistics (as of Fall 2022) at the University of Mi
 chigan.\n\nI work in the fields of Bayesian statistics\, machine learning\
 , and causal inference. I also work on algorithmic fairness and reinforcem
 ent learning. My research interests lie in the intersection of theory and 
 applications.\n\nPreviously\, I was a postdoctoral researcher with Profess
 or Michael Jordan at the University of California\, Berkeley. I completed 
 my Ph.D. in Statistics at Columbia University\, advised by Professor David
  Blei\, and my B.Sc. in mathematics and computer science at the Hong Kong 
 University of Science and Technology.\n\n
LOCATION:https://eng-cam.zoom.us/j/86187336902?pwd=eUtxZGsvandXQ1dXKzJkWXd
 lTFpqQT09
END:VEVENT
END:VCALENDAR