BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Learning-Algorithms from Bayesian Principles - Emti Khan\, RIKEN c
 enter for Advanced Intelligence Project
DTSTART:20190924T090000Z
DTEND:20190924T100000Z
UID:TALK130606@talks.cam.ac.uk
CONTACT:Dr R.E. Turner
DESCRIPTION:In machine learning\, new learning algorithms are designed by 
 borrowing ideas from optimization and statistics followed by an extensive 
 empirical efforts to make them practical. However\, there is a lack of und
 erlying principles to guide this process. I will present a stochastic lear
 ning algorithm derived from Bayesian principle. Using this algorithm\, we 
 can obtain a range of existing algorithms: from classical methods such as 
 least-squares\, Newton's method\, and Kalman filter to new deep-learning a
 lgorithms such as RMSprop and Adam. Surprisingly\, using the same principl
 es\, new algorithms can be naturally obtained even for the challenging lea
 rning tasks such as online learning\, continual learning\, and reinforceme
 nt learning. This talk will summarize recent works and outline future dire
 ctions on how this principle can be used to make algorithms that mimic the
  learning behaviour of living beings.
LOCATION:Engineering Department\, CBL Room BE-438.
END:VEVENT
END:VCALENDAR