Home
Reading
Searching
Subscribe
Sponsors
Statistics
Posting
Contact
Spam
Lists
Links
About
Hosting
Filtering
Features Download
Marketing
Archives
FAQ
Blog
 
Gmane
From: Sergey Lisitsyn <lisitsyn.s.o-Re5JQEeQqe8AvxtiuMwx3w <at> public.gmane.org>
Subject: GSoC Introduction - Domain adaptation and multitask learning algorithms
Newsgroups: gmane.comp.ai.machine-learning.shogun
Date: Tuesday 24th April 2012 07:11:32 UTC (over 4 years ago)
Dear all,

I hope I am known person there as I have been developing Shogun for a while
already :)

This year I am going to work on transfer learning algorithms, essentially
domain adaptation and multitask learning ones under guidance of Christian
Widmer. For the moment we have come with some preliminary plan including
following features:

   - Various tree based L1/L2 norm regression techniques ported from the
   SLEP package [1]
   - Domain adaptation and multitask learning multiclass support vector
   machines [2,5] (this is actually a field my mentor Christian is expert
in)
   - Structural correspondence learning [3]
   - Manifold regularized multitask learning [6]


This year GSoC gives me an opportunity to work full-time on our marvellous
Shogun toolbox and I'd like to thank Soeren and Christian for their
support. I also would like to thank all the GSoC candidates for carrying
out a few really great improvements during selection period. I am happy to
see our team is gradually expanding.

Cheers,
Sergey

[1] http://www.public.asu.edu/~jye02/Software/SLEP/
[2] http://rslab.disi.unitn.it/papers/R82-PAMI.pdf
[3] http://john.blitzer.com/papers/emnlp06.pdf
[4]
http://www.eecs.berkeley.edu/~arostami/papers/da_workshop/DAnips2011_Submitted.pdf
[5]
http://www.fml.tuebingen.mpg.de/raetsch/members/cwidmer/publications/pdfs/thesis.pdf
[6] http://www.umiacs.umd.edu/~arvinda/mysite/papers/nipsMTL.pdf
 
CD: 3ms