Learning Subspace Kernels for Classification

Jianhui Chen, Shuiwang Ji, Betul Ceran, Qi Li, Mingrui Wu, and Jieping Ye

　

---

Full Paper (PDF)

This paper is accepted by SIGKDD 2008.

---

Implementation Codes (Matlab)

The Subspace Kernel Learning Algorithms are implemented in Matlab 2007b. LIBSVM (MATLAB interface) is employed as the performance evaluation tools; you can download LIBSVM files from http://www.csie.ntu.edu.tw/~cjlin/libsvm/, complies them and then put them in the appropriate directory. MOSEK is employed as the (quadratical) optimization problems solver; you can download MOSEK from http://www.mosek.com/, and then apply a trial license. Detailed information about the codes is available in the README file.

---

Data Sets

Two types of data sets (UCI and Gene expression) are used in our experiments. The UCI data sets are available from http://archive.ics.uci.edu/ml/. The Gene expression data sets are available from http://www.gems-system.org/. In our experiments, all data sets are normalized to unit length.

---

If you have any questions or comments regarding the codes, please feel free to contact: Jianhui.Chen@asu.edu.