Description:

The following features are implemented: 1. Normalize sample data; 2. Search optimal C-gamma pairs for further training and prediction; 3. Train sample data and build hyper-sphere classification model; 4. Predict unknown-class sample data by using hyper-sphere model gotten from training process; 5. Cache kernel matrix for training of large scale sample data 6. Can run all processes by Ant script one-stop, it's very wonderful; 7. Count the executive time of all operations above.

Now this software only has Chinese documents about algorithm and its usage.If someone is interested in this, you can complete the English version. Welcome anyone to modify and improve this software! :) If you want to join us to improve HSSVM , please check SVN address: http://sourceforge.net/projects/hssvm/develop

Changes to previous version:

1. update the algorithm document;
2. add feature of writing prediction results to files when running all

BibTeX Entry: Download

Corresponding Paper BibTeX Entry: Download

Supported Operating Systems: Linux, Windows, Unix

Data Formats: Txt

Tags: Svm, Algorithm, Hypersphere, Multiclassification, Smo

Archive: download here

Comments

Patrick Lee (on March 31, 2010, 08:35:15)

just have a see~! O(?_?)O.lol:

Leave a comment

You must be logged in to post comments.