Skip to content

D
dlib
Commit 1650d8e9 authored by Davis King's avatar Davis King
Browse files
Options

Simplified the code for reduced() and reduced2() by making them use the 

kernel_matrix() and fill_lisf() funcions instead of the many for loops
they previously had.  In particular, using fill_lisf() makes the reducers
generally a lot faster and a little more accurate.

--HG--
extra : convert_revision : svn%3Afdd8eb12-d10e-0410-9acb-85c331704f74/trunk%403680
parent 7ef8e93b
Show whitespace changes
Inline Side-by-side
Showing with 17 additions and 84 deletions
dlib/svm/reduced.h
View file @ 1650d8e9
......@@ -80,53 +80,27 @@ namespace dlib
) const
{
// get the decision function object we are going to try and approximate
const decision_function<kernel_type> dec_funct = trainer.train(x,y);
const decision_function<kernel_type>& dec_funct = trainer.train(x,y);
// now find a linearly independent subset of the training points of num_bv points.
linearly_independent_subset_finder<kernel_type> lisf(dec_funct.kernel_function, num_bv);
for (long i = 0; i < x.nr(); ++i)
{
lisf.add(x(i));
}
fill_lisf(lisf, x);
// make num be the number of points in the lisf object. Just do this so we don't have
// to write out lisf.dictionary_size() all over the place.
const long num = lisf.dictionary_size();
// The next few blocks of code just find the best weights with which to approximate
// The next few statements just find the best weights with which to approximate
// the dec_funct object with the smaller set of vectors in the lisf dictionary. This
// is really just a simple application of some linear algebra. For the details
// see page 554 of Learning with kernels by Scholkopf and Smola where they talk
// about "Optimal Expansion Coefficients."
matrix<scalar_type, 0, 0, mem_manager_type> K_inv(num, num);
matrix<scalar_type, 0, 0, mem_manager_type> K(num, dec_funct.alpha.size());
const kernel_type kernel(dec_funct.kernel_function);
const kernel_type kern(dec_funct.kernel_function);
for (long r = 0; r < K_inv.nr(); ++r)
{
for (long c = 0; c < K_inv.nc(); ++c)
{
K_inv(r,c) = kernel(lisf[r], lisf[c]);
}
}
K_inv = pinv(K_inv);
matrix<scalar_type,0,1,mem_manager_type> alpha;
alpha = lisf.get_inv_kernel_marix()*(kernel_matrix(kern,lisf.get_dictionary(),dec_funct.basis_vectors)*dec_funct.alpha);
for (long r = 0; r < K.nr(); ++r)
{
for (long c = 0; c < K.nc(); ++c)
{
K(r,c) = kernel(lisf[r], dec_funct.basis_vectors(c));
}
}
// Now we compute the approximate decision function. Note that the weights come out
// of the expression K_inv*K*dec_funct.alpha.
decision_function<kernel_type> new_df(K_inv*K*dec_funct.alpha,
decision_function<kernel_type> new_df(alpha,
0,
kernel,
kern,
lisf.get_dictionary());
// now we have to figure out what the new bias should be. It might be a little
......@@ -492,49 +466,24 @@ namespace dlib
) const
{
// get the decision function object we are going to try and approximate
const decision_function<kernel_type> dec_funct = trainer.train(x,y);
const decision_function<kernel_type>& dec_funct = trainer.train(x,y);
// now find a linearly independent subset of the training points of num_bv points.
linearly_independent_subset_finder<kernel_type> lisf(dec_funct.kernel_function, num_bv);
for (long i = 0; i < x.nr(); ++i)
{
lisf.add(x(i));
}
fill_lisf(lisf,x);
// make num be the number of points in the lisf object. Just do this so we don't have
// to write out lisf.dictionary_size() all over the place.
const long num = lisf.dictionary_size();
// The next few blocks of code just find the best weights with which to approximate
// The next few statements just find the best weights with which to approximate
// the dec_funct object with the smaller set of vectors in the lisf dictionary. This
// is really just a simple application of some linear algebra. For the details
// see page 554 of Learning with kernels by Scholkopf and Smola where they talk
// about "Optimal Expansion Coefficients."
matrix<scalar_type, 0, 0, mem_manager_type> K_inv(num, num);
matrix<scalar_type, 0, 0, mem_manager_type> K(num, dec_funct.alpha.size());
const kernel_type kernel(dec_funct.kernel_function);
for (long r = 0; r < K_inv.nr(); ++r)
{
for (long c = 0; c < K_inv.nc(); ++c)
{
K_inv(r,c) = kernel(lisf[r], lisf[c]);
}
}
K_inv = pinv(K_inv);
const kernel_type kern(dec_funct.kernel_function);
for (long r = 0; r < K.nr(); ++r)
{
for (long c = 0; c < K.nc(); ++c)
{
K(r,c) = kernel(lisf[r], dec_funct.basis_vectors(c));
}
}
matrix<scalar_type,0,1,mem_manager_type> beta;
// Now we compute the fist approximate decision function.
matrix<scalar_type,0,1,mem_manager_type> beta(K_inv*K*dec_funct.alpha);
beta = lisf.get_inv_kernel_marix()*(kernel_matrix(kern,lisf.get_dictionary(),dec_funct.basis_vectors)*dec_funct.alpha);
matrix<sample_type,0,1,mem_manager_type> out_vectors(lisf.get_dictionary());
......@@ -554,29 +503,13 @@ namespace dlib
// out_vectors matrices
obj.vector_to_state(opt_starting_point);
// Do a final reoptimization of beta just to make sure it is optimal given the new
// set of basis vectors.
for (long r = 0; r < K_inv.nr(); ++r)
{
for (long c = 0; c < K_inv.nc(); ++c)
{
K_inv(r,c) = kernel(out_vectors(r), out_vectors(c));
}
}
K_inv = pinv(K_inv);
for (long r = 0; r < K.nr(); ++r)
{
for (long c = 0; c < K.nc(); ++c)
{
K(r,c) = kernel(out_vectors(r), dec_funct.basis_vectors(c));
}
}
beta = pinv(kernel_matrix(kern,out_vectors))*(kernel_matrix(kern,out_vectors,dec_funct.basis_vectors)*dec_funct.alpha);
decision_function<kernel_type> new_df(K_inv*K*dec_funct.alpha,
decision_function<kernel_type> new_df(beta,
0,
kernel,
kern,
out_vectors);
// now we have to figure out what the new bias should be. It might be a little
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment