| LIN2QUAD | |
Merges linear rule and quadratic mapping.
Synopsis:
quad_model = lin2quad(lin_model)
Description:
It recomputes parameters of input linear classifier
onto parameters of output quadratic classisifier.
The linear classifier is assumed to be trained on
the n-dimensional data quadraticaly mapped (see 'help qmap')
into the (n*(n+3)/2)-dimensional feature space. The linear
classifier in the feature space appears as the quadratic
classifier in the original data space.
Description:
orig_data = load('vltava');
map_data = qmap(orig_data);
lin_model = perceptron(map_data);
quad_model = lin2quad(lin_model);
figure; ppatterns(orig_data);
pboundary(quad_model);
See also
QUACLASS, QMAP
Modifications:
09-jun-2004, VF
17-may-2004, VF