BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:talk of Professor René Vidal (University of Pennsylvania)
DTSTART:20230207T140000
DTEND:20230207T150000
DTSTAMP:20260916T194623Z
UID:4edf6ed2d36c7a1fb936507bccf286cfba7cd3e5ee81b9fb81f1a08a
CATEGORIES:Conferences - Seminars
DESCRIPTION:Professor René Vidal\nTITLE: Explainable AI via Semantic Info
 rmation Pursuit\n\nABSTRACT: There is a significant interest in developing
  ML algorithms whose final predictions can be explained in terms understan
 dable to a human.  Providing such an “explanation” of the reasoning p
 rocess in domain-specific terms can be crucial for the adoption of ML algo
 rithms in risk-sensitive domains such as healthcare. This has motivated a 
 number of approaches that seek to provide explanations for existing ML alg
 orithms in a post-hoc manner.  However\, many of these approaches have be
 en widely criticized for a variety of reasons and no clear methodology exi
 sts in the field for developing ML algorithms whose predictions are readil
 y understandable by humans.  To address this challenge\, we develop a me
 thod for constructing high performance ML algorithms which are “explaina
 ble by design”. Namely\, our method makes its prediction by asking a se
 quence of domain- and task-specific yes/no queries about the data (akin to
  the game “20 questions”)\, each having a clear interpretation to the 
 end-user. We then minimize the expected number of queries needed for accur
 ate prediction on any given input. This allows for human interpretable und
 erstanding of the prediction process by construction\, as the questions wh
 ich form the basis for the prediction are specified by the user as interpr
 etable concepts about the data. Experiments on vision and NLP tasks demon
 strate the efficacy of our approach and its superiority over post-hoc expl
 anations. Joint work with Aditya Chattopadhyay\, Stewart Slocum\, Benjamin
  Haeffele and Donald Geman.\n\nSHORT BIOSKETCH: René Vidal is the Rachle
 ff Penn Integrates Knowledge University Professor in the Departments of El
 ectrical and Systems Engineering and Radiology and the inaugural Director 
 of the Institute for Data Engineering and Science (IDEAS) at University of
  Pennsylvania. He is also an Amazon Scholar\, a Chief Scientist at NORCE\,
  Associate Editor in Chief of TPAMI and the director of the NSF-Simons Col
 laboration on the Mathematical Foundations of Deep Learning and the NSF TR
 IPODS Institute on the Foundations of Graph and Deep Learning. His current
  research focuses on the foundations of deep learning and its applications
  in computer vision and biomedical data science. He is an ACM Fellow\, AIM
 BE Fellow\, IEEE Fellow\, IAPR Fellow and Sloan Fellow\, and has received 
 numerous awards for his work\, including the IEEE Edward J. McCluskey Tech
 nical Achievement Award\, D’Alembert Faculty Award\, J.K. Aggarwal Prize
 \, ONR Young Investigator Award\, NSF CAREER Award as well as best paper a
 wards in machine learning\, computer vision\, controls\, and medical robot
 ics.\n \n\nProfessor René Vidal © René Vidal \n\n\n\n\n 
LOCATION:ELA 2 https://plan.epfl.ch/?room==ELA%202 https://epfl.zoom.us/j/
 65819640839
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR