BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Talk of Dr Alessandro Rudi (INRIA and École Normale Supérieure)
DTSTART:20191122T111500
DTEND:20191122T130000
DTSTAMP:20260916T032117Z
UID:f0aea792067dc70968313a2ad5de557139eac32020074d57791e0e6d
CATEGORIES:Conferences - Seminars
DESCRIPTION: Dr Alessandro Rudi\nTitle:\nScaling up optimal kernel method
 s for large scale machine learning\n\nAbstract:\nKernel methods provide a 
 principled way to perform non linear\, nonparametric learning. They rely o
 n solid functional analytic foundations and enjoy optimal statistical prop
 erties. However\, at least in their basic form\, they have limited applica
 bility in large scale scenarios because of stringent computational require
 ments in terms of time and especially memory. In this talk we take a subst
 antial step in scaling up kernel methods analyzing novel algorithms techni
 ques that allow to efficiently process millions of points. The algorithms 
 are derived combining several algorithmic principles\, namely stochastic s
 ubsampling\, iterative solvers and preconditioning. Our theoretical analys
 is shows that optimal statistical accuracy is achieved requiring essential
 ly O(n) memory and O(n sqrt(n)) time. An extensive experimental analysis o
 n large scale datasets shows that\, even with a single machine\, the analy
 zed approach outperforms previous state of the art solutions\, which explo
 it parallel/distributed architectures.\n\nBIo:\n Alessandro Rudi is Rese
 archer at INRIA and École Normale Supérieure\, Paris. He received his P
 hD in 2014 from the University of Genova\, after being a visiting student 
 at the Center for Biological and Computational Learning at MIT. Between 2
 014 and 2017 he has been a postdoctoral fellow at Laboratory of Computatio
 nal and Statistical Learning at Italian Institute of Technology and Univer
 sity of Genova.
LOCATION:SG 0213 https://plan.epfl.ch/?room==SG%200213
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR