Optimisation of the Parzen classifier
[W,H] = PARZENC(A,H)
Computation of the optimum smoothing parameter H for the Parzen classifier between the classes in the dataset A. The leave-one-out Lissack && Fu estimate is used in the optimisation of H The final classifier is stored as a mapping in W. It may be converted into a classifier by W*CLASSC. PARZENC cannot be used for density estimation. The returned value of H, however, can be used in a the Parzen density estimator PARZENM.
The optimisation of H may be stopped prematurely by PRTIME.
In case smoothing H is specified, no learning is performed, just the discriminant W is produced for the given smoothing parameters H. Smoothing parameters may be scalar, vector of per-class parameters, or a matrix with individual smoothing for each class (rows) and feature directions (columns)
prex_density, for, densities, and, prex_parzen, for, differences, between,
PARZENC, PARZENDC and PARZENM, PRTIME
T. Lissack and K.S. Fu, Error estimation in pattern recognition via L-distance between posterior density functions, IEEE Trans. Inform. Theory, vol. 22, pp. 34-45, 1976.
datasets, mappings, parzen_map, parzenml, parzendc, parzenm, classc, prex_parzen,