BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Large Graph Limits of Learning Algorithms - Andrew Stuart (Univers
 ity of Warwick)
DTSTART:20180410T123000Z
DTEND:20180410T133000Z
UID:TALK103594@talks.cam.ac.uk
CONTACT:INI IT
DESCRIPTION:Many problems in machine learning require the classification&n
 bsp\;of high dimensional data. One methodology to approach such&nbsp\;prob
 lems is to construct a graph whose vertices are identified with data point
 s\,&nbsp\;with edges weighted according to some measure of affinity betwee
 n  the data points. Algorithms such as spectral clustering\, probit&nbsp\;
 classification and the Bayesian level set method can all be applied  in th
 is setting. The goal of the talk is to describe these algorithms&nbsp\;for
  classification\, and analyze them in the limit of large data  sets. Doing
  so leads to interesting problems in the calculus of&nbsp\;variations\, Ba
 yesian inverse problems and in&nbsp\;Monte Carlo Markov Chain\, all of whi
 ch will be highlighted in the&nbsp\;talk. These limiting problems give ins
 ight into the structure of&nbsp\;the classification problem\, and algorith
 ms for it.  &nbsp\;  &nbsp\;  <br><br>Collaboration with:  &nbsp\;  <br>An
 drea Bertozzi (UCLA)  <br>Michael Luo (UCLA)  <br>Kostas Zygalakis (Edinbu
 rgh)  <br><a target="_blank" rel="nofollow" href="https://arxiv.org/abs/17
 03.08816">https://arxiv.org/abs/1703.08816</a>  &nbsp\; <br> and  &nbsp\; 
  <br>Matt Dunlop (Caltech)  <br>Dejan Slepcev (CMU)  <br>Matt Thorpe (Camb
 ridge)  <br>(forthcoming paper)  <br>
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR