OpenKernel Background Material

Kernel methods (such as SVM, perceptron, kernel PCA) are widely used in machine learning. Applying these methods in text and language processing or bioinformatics requires to be able to define kernels on sequences or distributions of sequences. Rational kernels are a general framework for defining such kernels. Commonly used kernels on sequences such as n-gram kernels, gappy n-gram kernels, mismatch kernels, and Hausler's convolution kernels are special cases of rational kernels.

The notion of rational kernel was introduced by: Corinna Cortes, Patrick Haffner and Mehryar Mohri, Rational Kernels: Theory and Algorithms, Journal of Machine Learning Research 5:1035--1062, 2004.

References

Rational kernels

• Corinna Cortes, Patrick Haffner and Mehryar Mohri, Rational Kernels: Theory and Algorithms, Journal of Machine Learning Research 5:1035--1062, 2004.

Kernels on string and sequences

• Huma Lodhi, Craig Saunders, John Shawe-Taylor, Nello Cristianini and Chris Watskins. Text Classification using String Kernels, Journal of Machine Learning Research 2:419-444, 2002.

• Christina S. Leslie, Eleazar Eskin, Adiel Cohen, Jason Weston and William Stafford Noble. Mismatch string kernels for discriminative protein classification. Bioinformatics 20(4):467-476, 2004.

• David Hausler. Convolution kernels on discrete structures. Technical Report UCSC-CRL-99-10, University of California at Santa Cruz, 1999.

Kernel methods

• Bernard Scholköpf and Alex Smola. Learning with kernels. MIT Press, 2002.

-- CyrilAllauzen - 08 Oct 2007