Home
Reading
Searching
Subscribe
Sponsors
Statistics
Posting
Contact
Spam
Lists
Links
About
Hosting
Filtering
Features Download
Marketing
Archives
FAQ
Blog
 
Gmane
From: =?UTF-8?B?TWljaGFsIFXFmWnEjcOhxZk=?= <uricar.michal-Re5JQEeQqe8AvxtiuMwx3w <at> public.gmane.org>
Subject: GSoC Introduction - Bundle method solver for SO learning
Newsgroups: gmane.comp.ai.machine-learning.shogun
Date: Tuesday 24th April 2012 13:06:49 UTC (over 4 years ago)
Hi everybody,

my name is Michal Uricar (my nickname on IRC is uricamic) and I am a first
year PhD student from Czech Technical university in Prague, Czech Republic.
I have been selected for the task of developing bundle methods solver for
SO learning mentored by Vojta Franc.

My work consists of three main parts:

   1. Implementation of the classical Bundle Method for SO learning [1][3]
   2. Integration of this algorithm to Shogun
   3. Experimental comparison of this algorithm with the state of the art
   structSVM [2] and other SO solvers already integrated in Shogun

Besides the contact with my mentor and other Shogun developers through both
mailing list and IRC channel, I plan closer collaboration with Fernando J.
Iglesias Garcia, GSoC student whose task is development of the generic SO
framework.

Finally, please let me express that I am really grateful for the
opportunity to participate in this wonderful project.

Best regards

Michal.

[1] C.H. Teo, S.V.N. Vishwanthan, A.J. Smola, Q.V. Le. Bundle Methods for
Regularized Risk
Minimization<http://jmlr.csail.mit.edu/papers/volume11/teo10a/teo10a.pdf>.
JMLR, 11(Jan):311-365, 2010.
[2] T.Joachims. StructSVM <http://svmlight.joachims.org/svm_struct.html>.
[3] C. Lemarechal, A.Nemirovskii, Y. Nesterov. New variants of bundle
methods. Mathematical Programming. 69, 111-147. 1995
 
CD: 3ms