Optimally regularised kernel Fisher discriminant analysis

K. Saadi, N. L. C. Talbot, G. C. Cawley

Research output: Contribution to conferencePaper

7 Citations (Scopus)


Mika et al. (1999) introduce a non-linear formulation of Fisher's linear discriminant, based the now familiar "kernel trick", demonstrating state-of-the-art performance on a wide range of real-world benchmark datasets. In this paper, we show that the usual regularisation parameter can be adjusted so as to minimise the leave-one-out cross-validation error with a computational complexity of only O(l2) operations, where l is the number of training patterns, rather than the O(l4) operations required for a naive implementation of the leave-one-out procedure. This procedure is then used to form a component of an efficient hierarchical model selection strategy where the regularisation parameter is optimised within the inner loop while the kernel parameters are optimised in the outer loop.
Original languageEnglish
Number of pages4
Publication statusPublished - Aug 2004
Event17th International Conference on Pattern Recognition - Cambridge, United Kingdom
Duration: 23 Aug 200426 Aug 2004


Conference17th International Conference on Pattern Recognition
Abbreviated titleICPR-2004
Country/TerritoryUnited Kingdom

Cite this