BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:The Minimum Description Length Principle and Machine Learning - Dr
  Yoshinari Takeishi\, Kyushu University
DTSTART:20260128T140000Z
DTEND:20260128T150000Z
UID:TALK239470@talks.cam.ac.uk
CONTACT:Prof. Ramji Venkataramanan
DESCRIPTION:The Minimum Description Length (MDL) principle states that goo
 d learning can be achieved by selecting the model that provides the shorte
 st description of the observed data. It is a key concept that bridges info
 rmation theory and machine learning\, enabling us to understand increasing
 ly important machine learning problems from an information-theoretic viewp
 oint. In this talk\, we first review methods for efficient lossless compre
 ssion of data generated from an unknown probability distribution (universa
 l coding)\, with a particular focus on two-stage (two-part) coding. We the
 n introduce the MDL estimator based on two-stage codes and explain how it 
 relates to standard learning formulations. Finally\, we present a theorem 
 by Barron and Cover that provides a generalization guarantee for this MDL 
 estimator\, thereby offering a rigorous mathematical justification for app
 lying the MDL principle in machine learning.
LOCATION:MR5\, CMS Pavilion A
END:VEVENT
END:VCALENDAR