Description:

LibSGDQN proposes an implementation of SGD-QN, a carefully designed quasi-Newton stochastic gradient descent solver for linear SVMs.

The SGD-QN algorithm is a stochastic gradient descent algorithm that makes careful use of second order information and splits the parameter update into independently scheduled components. Thanks to this design, SGD-QN iterates nearly as fast as a first order stochastic gradient descent but requires less iterations to achieve the same accuracy.

This algorithm is extensively described in the paper: "SGD-QN: Careful Quasi-Newton Stochastic Gradient Descent" by A. Bordes, L. Bottou and P. Gallinari published in Journal of Machine Learning Research: Special Topic on Large Scale Learning (2009).

Along with SGD-QN, this library proposes the implementation of two other online solvers for linear SVMs (also discussed in the JMLR paper). A script to re-run the experiments of this paper is also provided.

Changes to previous version:

small bug fix (thx nicolas ;)

BibTeX Entry: Download

Corresponding Paper BibTeX Entry: Download

Supported Operating Systems: Cygwin, Macosx, Unix

Data Formats: Svmlight

Tags: Large Scale, Support Vector Machines, Stochastic Gradient Descent

Archive: download here

Other available revisons

Version	Changelog	Date
1.1	small bug fix (thx nicolas ;)	July 2, 2009, 15:02:44
1.0	Initial Announcement on mloss.org.	June 24, 2009, 16:51:12

Comments

looya.B (on July 8, 2009, 07:41:42)

sorry, there is nothing in my download files. what happened?

Antoine Bordes (on July 8, 2009, 08:27:28)

Did you run the 'get_datat.sh' script in the 'data/' directory? This downloads the datafiles (and might take quite a while ... ) It's working ok for me.

Leave a comment

You must be logged in to post comments.