| |
RSC_l0 reduced set selection
|
a = rss_l0(alg,hyper)
generates a rss object, using the l0 norm
hyperparameters:
child=svm algorithm worked on
reoptimize_b=1 recalculate the threshold b0
a.lambda=0 regularizer for selection
set to -1 for automatic selection
reoptimize=1
tolerance=0.05 tolerance in ||w-w*||^2
max_loops=40 limit on loops
model:
alpha new alphas for rs-vectors
Xsv rs vectors
b0 the threshold
w2 final value of ||w*-w^2||^2, set to -1 to calculate
stats:
w2=0 final value of ||w-w*||^2
res=[] results on a separate test set
dtst=[] separate test set
test_on=0 iterations to test on
methods:
train constructs a reduced set, returns trained rs-machine
test tests new rs-machine on supplied data
example:
d=gen(toy2d('2circles','l=100'));
[r,a]=train(svm({kernel('rbf',1),'C=10000','alpha_cutoff=1e-2'}),d);
[r,a2]=train(rss_l0(a,'lambda=1e-2'),d);
test(a2,d,loss)
author: goekhan bakir, jason weston
reference: fast binary and multi-output rss, 2004