BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Active Learning - Austin Tripp and Erik Daxberger  (University of 
 Cambridge)
DTSTART:20200624T100000Z
DTEND:20200624T113000Z
UID:TALK149725@talks.cam.ac.uk
CONTACT:75379
DESCRIPTION:In many real-world problems such as medical diagnosis\, labell
 ing data is expensive and labelled datasets are scarce. Active learning de
 scribes a set of machine learning techniques that choose their own trainin
 g data. The key idea behind active learning is to achieve greater accuracy
  with fewer training labels by allowing the ML algorithm to choose the dat
 a from which it learns. To this end\, the active learning algorithm poses 
 queries (i.e. unlabelled data instances) to be labelled by an oracle (e.g.
 \, a human annotator).\n\nIn this reading group session\, we aim to provid
 e a broad overview over both the foundations and frontiers of active learn
 ing. We will first define active learning\, contrast it against other form
 ulations of machine learning\, and describe the standard techniques used t
 o select queries. We will then survey recent advances in active learning\,
  often in the context of (Bayesian) deep neural networks. We hope to concl
 ude with an open-ended discussion on active learning.
LOCATION:https://meet.google.com/xom-namz-rzv
END:VEVENT
END:VCALENDAR