GRACS

This is an old revision of the document!

Location: ACES 3.116 Time: Wednesdays, 10-11, every other week. Other reading groups: Reading Groups

Theme

Graph Laplacians and large scale data mining.

Papers to read

Paper	Amar Subramanya and Jeff A. Bilmes. Entropic Graph Regularization in Non-Parametric Semi-Supervised Classification. In Neural Information Processing Society (NIPS), Vancouver, Canada, December 2009..
Presenter
Date	17 Mar 2010.
Paper	Label Propagation and Quadratic Criterion, Yoshua Bengio, Olivier Delalleau, Nicolas Le Roux [Chapter of the book Semi-supervised learning by O. Chapelle, Bernhard Schölkopf and Alexander Zien]
Presenter	Donghyuk Shin
Date	3 Mar 2010.
Paper	Semi-supervised Learning in Gigantic Image Collections Rob Fergus, Yair Weiss and Antonio Torralba, NIPS 2009..
Presenter	Nagarajan Natarajan
Date	17th Feb 2010.

—-

Paper	http://www.cs.cmu.edu/~skumar/ens.pdf Ensemble Nystrom Method, Sanjiv Kumar, Mehryar Mohri and Ameet Talwalkar
Proposer	Zhengdong Lu
Why?	similar topic as the two other papers from NIPS09

Paper	http://www.cbrc.jp/~tsuda/pdf/icml05.pdf Propagating Distributions on a Hypergrap by Dual Information Regularization, Koji Tsuda. ICML05
Proposer	Zhengdong Lu
Why?	A paper on KL-based objective for semi-supervised learning paper. Relevant to the Rob Fergus paper, actually cited by them. I read it years ago and found it interesting.

Paper	Mark Herbster, Massimiliano Pontil, Sergio Rojas Galeano: Fast Prediction on a Tree. 657-664 http://books.nips.cc/papers/files/nips21/NIPS2008_0824.pdf
Proposer	Inderjit Dhillon
Why?

Paper	N. Halko, P.G. Martinsson, and J.A. Tropp: Finding structure with randomness PAPER: [amath.colorado.edu/faculty/martinss/Pubs/2009_HMT_random_review.pdf] SLIDES: [www2.cse.iitk.ac.in/~fsttcs/2009/wapmds/slides/JoelTropp.pdf]
Proposer	Berkant Savas
Why?	The paper may be too long (81 pages) to print out but the ideas are very simple. Check the slides first.