Abstract

We present a fast training algorithm for the kernel Fisher discriminant classifier. It uses a greedy approximation technique and has an empirical scaling behavior which improves upon the state of the art by more than an order of magnitude, thus rendering the kernel Fisher algorithm a viable option also for large datasets. 1

Citations

1523	Matrix Computations - Golub, Loan - 1983
936	A training algorithm for optimal margin classifiers - Boser, Guyon, et al. - 1992
716	The use of multiple measurements in taxonomic problems - Fisher - 1936
231	Fisher discriminant analysis with kernels - Mika, Rätsch, et al. - 1999
139	Sparse greedy matrix approximation for machine learning - Smola, Schölkopf - 2000
93	Sparse greedy gaussian process regression - Smola, Bartlett - 2001
32	A column generation algorithm for boosting - Bennett, Demiriz, et al. - 2000
32	Invariant feature extraction and classification in kernel spaces - Mika, Rätsch, et al. - 2000
21	ªSparse Regression Ensembles in Infinite and Finite Hypothesis Spaces,º - RaÈtsch, Demiriz, et al. - 2002
10	Kernel methods for extracting local image semantics - Bradshaw, Schölkopf, et al. - 2001
5	Invariant feature extraction and classi in kernel spaces - Mika, Ratsch, et al. - 2000
2	Sparse regression ensembles in in and hypothesis spaces - Ratsch, Demiriz, et al. - 2000
1	3rd edition - Mercer - 1996
1	A 209:415– 446 - London - 1909