BEGIN:VCALENDAR
VERSION:2.0
CALSCALE:GREGORIAN
PRODID:UW-Madison-Physics-Events
BEGIN:VEVENT
SEQUENCE:4
UID:UW-Physics-Event-6641
DTSTART:20211201T170000Z
DTEND:20211201T181500Z
DTSTAMP:20260414T191915Z
LAST-MODIFIED:20211120T030207Z
LOCATION:Chamberlin 5280 (Zoom link also available for online particip
 ants who signed up on our mailing list)
SUMMARY:Uncovering the Unknowns of Deep Neural Networks: Challenges an
 d Opportunities\, Physics ∩ ML Seminar\, Sharon Li\, University of W
 isconsin-Madison
DESCRIPTION:The real world is open and full of unknowns\, presenting s
 ignificant challenges for machine learning (ML) systems that must reli
 ably handle diverse\, and sometimes anomalous inputs. Out-of-distribut
 ion (OOD) uncertainty arises when a machine learning model sees a test
 -time input that differs from its training data\, and thus should not 
 be predicted by the model. As ML is used for more safety-critical doma
 ins\, the ability to handle out-of-distribution data are central in bu
 ilding open-world learning systems. In this talk\, I will talk about c
 hallenges\, methods\, and opportunities on uncovering the unknowns of 
 deep neural networks for reliable decision-making in an open world. 
URL:https://www.physics.wisc.edu/events/?id=6641
END:VEVENT
END:VCALENDAR