PLearn 0.1
|
#include <AdditiveNormalizationKernel.h>
Public Member Functions | |
AdditiveNormalizationKernel () | |
Default constructor. | |
AdditiveNormalizationKernel (Ker the_source, bool remove_bias=false, bool remove_bias_in_evaluate=false, bool double_centering=false) | |
Created from an existing kernel. | |
virtual void | build () |
Simply calls inherited::build() then build_(). | |
virtual void | makeDeepCopyFromShallowCopy (CopiesMap &copies) |
Transforms a shallow copy into a deep copy. | |
virtual string | classname () const |
virtual OptionList & | getOptionList () const |
virtual OptionMap & | getOptionMap () const |
virtual RemoteMethodMap & | getRemoteMethodMap () const |
virtual AdditiveNormalizationKernel * | deepCopy (CopiesMap &copies) const |
virtual real | evaluate (const Vec &x1, const Vec &x2) const |
Overridden. | |
virtual real | evaluate_i_j (int i, int j) const |
returns evaluate(data(i),data(j)) | |
virtual real | evaluate_i_x (int i, const Vec &x, real squared_norm_of_x=-1) const |
Return evaluate(data(i),x). | |
virtual real | evaluate_x_i (const Vec &x, int i, real squared_norm_of_x=-1) const |
returns evaluate(x,data(i)) [default version calls evaluate_i_x if kernel is_symmetric] | |
virtual real | evaluate_i_x_again (int i, const Vec &x, real squared_norm_of_x=-1, bool first_time=false) const |
Return evaluate(data(i),x), where x is the same as in the precedent call to this same function (except if 'first_time' is true). | |
virtual real | evaluate_x_i_again (const Vec &x, int i, real squared_norm_of_x=-1, bool first_time=false) const |
virtual void | computeGramMatrix (Mat K) const |
Call evaluate_i_j to fill each of the entries (i,j) of symmetric matrix K. | |
virtual void | setDataForKernelMatrix (VMat the_data) |
** Subclasses may override these methods to provide efficient kernel matrix access ** | |
Static Public Member Functions | |
static string | _classname_ () |
static OptionList & | _getOptionList_ () |
static RemoteMethodMap & | _getRemoteMethodMap_ () |
static Object * | _new_instance_for_typemap_ () |
static bool | _isa_ (const Object *o) |
static void | _static_initialize_ () |
static const PPath & | declaringFile () |
Public Attributes | |
bool | data_will_change |
bool | double_centering |
bool | remove_bias |
bool | remove_bias_in_evaluate |
Static Public Attributes | |
static StaticInitializer | _static_initializer_ |
Protected Member Functions | |
real | computeAverage (const Vec &x, bool on_row, real squared_norm_of_x=-1) const |
Return the average of K(x,x_i) or K(x_i,x), depending on the value of 'on_row' (true or false, respectively). | |
Static Protected Member Functions | |
static void | declareOptions (OptionList &ol) |
Declares this class' options. | |
Protected Attributes | |
Vec | average_col |
Vec | average_row |
real | total_average |
real | total_average_unbiased |
real | avg_evaluate_i_x_again |
The last average computed in evaluate_i_x_again(). | |
real | avg_evaluate_x_i_again |
The last average computed in evaluate_x_i_again(). | |
real | factor |
A multiplicative factor to scale the result (1 or -1/2). | |
Private Types | |
typedef SourceKernel | inherited |
Private Member Functions | |
void | build_ () |
This does the actual building. | |
Private Attributes | |
Vec | all_k_x |
Used to store the values of the source kernel. |
Definition at line 52 of file AdditiveNormalizationKernel.h.
typedef SourceKernel PLearn::AdditiveNormalizationKernel::inherited [private] |
Reimplemented from PLearn::SourceKernel.
Definition at line 57 of file AdditiveNormalizationKernel.h.
PLearn::AdditiveNormalizationKernel::AdditiveNormalizationKernel | ( | ) |
Default constructor.
Definition at line 52 of file AdditiveNormalizationKernel.cc.
: total_average(0.), total_average_unbiased(0.), data_will_change(false), double_centering(false), remove_bias(false), remove_bias_in_evaluate(false) { }
PLearn::AdditiveNormalizationKernel::AdditiveNormalizationKernel | ( | Ker | the_source, |
bool | remove_bias = false , |
||
bool | remove_bias_in_evaluate = false , |
||
bool | double_centering = false |
||
) |
Created from an existing kernel.
Definition at line 62 of file AdditiveNormalizationKernel.cc.
: total_average(0.), total_average_unbiased(0.), data_will_change(false), double_centering(the_double_centering), remove_bias(the_remove_bias), remove_bias_in_evaluate(the_remove_bias_in_evaluate) { source_kernel = the_source; build(); }
string PLearn::AdditiveNormalizationKernel::_classname_ | ( | ) | [static] |
Reimplemented from PLearn::SourceKernel.
Definition at line 85 of file AdditiveNormalizationKernel.cc.
OptionList & PLearn::AdditiveNormalizationKernel::_getOptionList_ | ( | ) | [static] |
Reimplemented from PLearn::SourceKernel.
Definition at line 85 of file AdditiveNormalizationKernel.cc.
RemoteMethodMap & PLearn::AdditiveNormalizationKernel::_getRemoteMethodMap_ | ( | ) | [static] |
Reimplemented from PLearn::SourceKernel.
Definition at line 85 of file AdditiveNormalizationKernel.cc.
Reimplemented from PLearn::SourceKernel.
Definition at line 85 of file AdditiveNormalizationKernel.cc.
Object * PLearn::AdditiveNormalizationKernel::_new_instance_for_typemap_ | ( | ) | [static] |
Reimplemented from PLearn::SourceKernel.
Definition at line 85 of file AdditiveNormalizationKernel.cc.
StaticInitializer AdditiveNormalizationKernel::_static_initializer_ & PLearn::AdditiveNormalizationKernel::_static_initialize_ | ( | ) | [static] |
Reimplemented from PLearn::SourceKernel.
Definition at line 85 of file AdditiveNormalizationKernel.cc.
void PLearn::AdditiveNormalizationKernel::build | ( | ) | [virtual] |
Simply calls inherited::build() then build_().
Reimplemented from PLearn::SourceKernel.
Definition at line 131 of file AdditiveNormalizationKernel.cc.
References PLearn::SourceKernel::build(), and build_().
{ // ### Nothing to add here, simply calls build_ inherited::build(); build_(); }
void PLearn::AdditiveNormalizationKernel::build_ | ( | ) | [private] |
This does the actual building.
Reimplemented from PLearn::SourceKernel.
Definition at line 141 of file AdditiveNormalizationKernel.cc.
References double_centering, and factor.
Referenced by build().
{ // ### This method should do the real building of the object, // ### according to set 'options', in *any* situation. // ### Typical situations include: // ### - Initial building of an object from a few user-specified options // ### - Building of a "reloaded" object: i.e. from the complete set of all serialised options. // ### - Updating or "re-building" of an object after a few "tuning" options have been modified. // ### You should assume that the parent class' build_() has already been called. if (double_centering) factor = -0.5; else factor = 1; }
string PLearn::AdditiveNormalizationKernel::classname | ( | ) | const [virtual] |
Reimplemented from PLearn::SourceKernel.
Definition at line 85 of file AdditiveNormalizationKernel.cc.
real PLearn::AdditiveNormalizationKernel::computeAverage | ( | const Vec & | x, |
bool | on_row, | ||
real | squared_norm_of_x = -1 |
||
) | const [inline, protected] |
Return the average of K(x,x_i) or K(x_i,x), depending on the value of 'on_row' (true or false, respectively).
Definition at line 159 of file AdditiveNormalizationKernel.cc.
References all_k_x, PLearn::Kernel::is_symmetric, PLearn::Kernel::n_examples, PLearn::TVec< T >::resize(), PLearn::SourceKernel::source_kernel, and PLearn::sum().
Referenced by evaluate(), evaluate_i_x(), evaluate_i_x_again(), evaluate_x_i(), and evaluate_x_i_again().
{ all_k_x.resize(n_examples); if (is_symmetric || !on_row) { source_kernel->evaluate_all_i_x(x, all_k_x, squared_norm_of_x); } else { source_kernel->evaluate_all_x_i(x, all_k_x, squared_norm_of_x); } return sum(all_k_x) / real(n_examples); }
void PLearn::AdditiveNormalizationKernel::computeGramMatrix | ( | Mat | K | ) | const [virtual] |
Call evaluate_i_j to fill each of the entries (i,j) of symmetric matrix K.
Reimplemented from PLearn::SourceKernel.
Definition at line 172 of file AdditiveNormalizationKernel.cc.
{ // Uses default Kernel implementation. Kernel::computeGramMatrix(K); }
void PLearn::AdditiveNormalizationKernel::declareOptions | ( | OptionList & | ol | ) | [static, protected] |
Declares this class' options.
Reimplemented from PLearn::SourceKernel.
Definition at line 90 of file AdditiveNormalizationKernel.cc.
References average_col, average_row, PLearn::OptionBase::buildoption, data_will_change, PLearn::declareOption(), PLearn::SourceKernel::declareOptions(), double_centering, PLearn::OptionBase::learntoption, remove_bias, remove_bias_in_evaluate, total_average, and total_average_unbiased.
{ // Build options. declareOption(ol, "double_centering", &AdditiveNormalizationKernel::double_centering, OptionBase::buildoption, "If set to 1, then the resulting kernel will be multiplied by -1/2,\n" "which corresponds to the double-centering formula."); declareOption(ol, "data_will_change", &AdditiveNormalizationKernel::data_will_change, OptionBase::buildoption, "If set to 1, then the Gram matrix will be always recomputed, even if\n" "it's not completely sure the data has changed."); declareOption(ol, "remove_bias", &AdditiveNormalizationKernel::remove_bias, OptionBase::buildoption, "If set to 1, then the bias induced by the K(x_i,x_i) will be removed.\n"); declareOption(ol, "remove_bias_in_evaluate", &AdditiveNormalizationKernel::remove_bias_in_evaluate, OptionBase::buildoption, "If set to 1, then the bias induced by the K(x_i,x_i) will be removed, but only when\n" "evaluating K(x,y) on test points (you don't need to do this if 'remove_bias' == 1)."); // Learnt options. declareOption(ol, "average_col", &AdditiveNormalizationKernel::average_col, OptionBase::learntoption, "The average of the underlying kernel over each column of the Gram matrix."); declareOption(ol, "average_row", &AdditiveNormalizationKernel::average_row, OptionBase::learntoption, "The average of the underlying kernel over each row of the Gram matrix."); declareOption(ol, "total_average_unbiased", &AdditiveNormalizationKernel::total_average_unbiased, OptionBase::learntoption, "The average of the underlying kernel over the whole Gram matrix, without\n" "the diagonal terms."); declareOption(ol, "total_average", &AdditiveNormalizationKernel::total_average, OptionBase::learntoption, "The average of the underlying kernel over the whole Gram matrix."); // Now call the parent class' declareOptions inherited::declareOptions(ol); }
static const PPath& PLearn::AdditiveNormalizationKernel::declaringFile | ( | ) | [inline, static] |
Reimplemented from PLearn::SourceKernel.
Definition at line 138 of file AdditiveNormalizationKernel.h.
AdditiveNormalizationKernel * PLearn::AdditiveNormalizationKernel::deepCopy | ( | CopiesMap & | copies | ) | const [virtual] |
Reimplemented from PLearn::SourceKernel.
Definition at line 85 of file AdditiveNormalizationKernel.cc.
real PLearn::AdditiveNormalizationKernel::evaluate | ( | const Vec & | x1, |
const Vec & | x2 | ||
) | const [virtual] |
Overridden.
Reimplemented from PLearn::SourceKernel.
Definition at line 180 of file AdditiveNormalizationKernel.cc.
References computeAverage(), factor, remove_bias, remove_bias_in_evaluate, PLearn::SourceKernel::source_kernel, total_average, and total_average_unbiased.
{ real avg_1 = computeAverage(x1, true); real avg_2 = computeAverage(x2, false); if (remove_bias || !remove_bias_in_evaluate) { // We can use the 'total_average'. return factor * (source_kernel->evaluate(x1, x2) - avg_1 - avg_2 + total_average); } else { // We need to use the 'total_average_unbiased'. return factor * (source_kernel->evaluate(x1, x2) - avg_1 - avg_2 + total_average_unbiased); } }
returns evaluate(data(i),data(j))
Reimplemented from PLearn::SourceKernel.
Definition at line 195 of file AdditiveNormalizationKernel.cc.
References average_col, average_row, factor, i, j, PLearn::SourceKernel::source_kernel, and total_average.
{ return factor * (source_kernel->evaluate_i_j(i,j) - average_row[i] - average_col[j] + total_average); }
real PLearn::AdditiveNormalizationKernel::evaluate_i_x | ( | int | i, |
const Vec & | x, | ||
real | squared_norm_of_x = -1 |
||
) | const [virtual] |
Return evaluate(data(i),x).
[squared_norm_of_x is just a hint that may allow to speed up computation if it is already known, but it's optional]
Reimplemented from PLearn::SourceKernel.
Definition at line 202 of file AdditiveNormalizationKernel.cc.
References average_row, computeAverage(), factor, i, PLearn::SourceKernel::source_kernel, and total_average.
{ return factor * (source_kernel->evaluate_i_x(i, x, squared_norm_of_x) - average_row[i] - computeAverage(x, false, squared_norm_of_x) + total_average); }
real PLearn::AdditiveNormalizationKernel::evaluate_i_x_again | ( | int | i, |
const Vec & | x, | ||
real | squared_norm_of_x = -1 , |
||
bool | first_time = false |
||
) | const [virtual] |
Return evaluate(data(i),x), where x is the same as in the precedent call to this same function (except if 'first_time' is true).
This can be used to speed up successive computations of K(x_i, x) (default version just calls evaluate_i_x).
Reimplemented from PLearn::Kernel.
Definition at line 210 of file AdditiveNormalizationKernel.cc.
References average_row, avg_evaluate_i_x_again, computeAverage(), factor, i, PLearn::SourceKernel::source_kernel, and total_average.
{ if (first_time) { avg_evaluate_i_x_again = computeAverage(x, false, squared_norm_of_x); } return factor * (source_kernel->evaluate_i_x_again(i, x, squared_norm_of_x, first_time) - average_row[i] - avg_evaluate_i_x_again + total_average); }
real PLearn::AdditiveNormalizationKernel::evaluate_x_i | ( | const Vec & | x, |
int | i, | ||
real | squared_norm_of_x = -1 |
||
) | const [virtual] |
returns evaluate(x,data(i)) [default version calls evaluate_i_x if kernel is_symmetric]
Reimplemented from PLearn::SourceKernel.
Definition at line 221 of file AdditiveNormalizationKernel.cc.
References average_col, computeAverage(), factor, i, PLearn::SourceKernel::source_kernel, and total_average.
{ return factor * (source_kernel->evaluate_x_i(x, i, squared_norm_of_x) - average_col[i] - computeAverage(x, true, squared_norm_of_x) + total_average); }
real PLearn::AdditiveNormalizationKernel::evaluate_x_i_again | ( | const Vec & | x, |
int | i, | ||
real | squared_norm_of_x = -1 , |
||
bool | first_time = false |
||
) | const [virtual] |
Reimplemented from PLearn::Kernel.
Definition at line 229 of file AdditiveNormalizationKernel.cc.
References average_col, avg_evaluate_x_i_again, computeAverage(), factor, i, PLearn::SourceKernel::source_kernel, and total_average.
{ if (first_time) { avg_evaluate_x_i_again = computeAverage(x, true, squared_norm_of_x); } return factor * (source_kernel->evaluate_x_i_again(x, i, squared_norm_of_x, first_time) - average_col[i] - avg_evaluate_x_i_again + total_average); }
OptionList & PLearn::AdditiveNormalizationKernel::getOptionList | ( | ) | const [virtual] |
Reimplemented from PLearn::SourceKernel.
Definition at line 85 of file AdditiveNormalizationKernel.cc.
OptionMap & PLearn::AdditiveNormalizationKernel::getOptionMap | ( | ) | const [virtual] |
Reimplemented from PLearn::SourceKernel.
Definition at line 85 of file AdditiveNormalizationKernel.cc.
RemoteMethodMap & PLearn::AdditiveNormalizationKernel::getRemoteMethodMap | ( | ) | const [virtual] |
Reimplemented from PLearn::SourceKernel.
Definition at line 85 of file AdditiveNormalizationKernel.cc.
void PLearn::AdditiveNormalizationKernel::makeDeepCopyFromShallowCopy | ( | CopiesMap & | copies | ) | [virtual] |
Transforms a shallow copy into a deep copy.
Reimplemented from PLearn::SourceKernel.
Definition at line 240 of file AdditiveNormalizationKernel.cc.
References all_k_x, average_col, average_row, PLearn::deepCopyField(), and PLearn::SourceKernel::makeDeepCopyFromShallowCopy().
{ inherited::makeDeepCopyFromShallowCopy(copies); deepCopyField(all_k_x, copies); deepCopyField(average_col, copies); deepCopyField(average_row, copies); }
void PLearn::AdditiveNormalizationKernel::setDataForKernelMatrix | ( | VMat | the_data | ) | [virtual] |
** Subclasses may override these methods to provide efficient kernel matrix access **
This method sets the data VMat that will be used to define the kernel matrix. It may precompute values from this that may later accelerate the evaluation of a kernel matrix element
Reimplemented from PLearn::SourceKernel.
Definition at line 251 of file AdditiveNormalizationKernel.cc.
References average_col, average_row, PLearn::Kernel::data, data_will_change, PLearn::TVec< T >::fill(), i, PLearn::Kernel::is_symmetric, j, PLearn::TVec< T >::length(), PLearn::VMat::length(), n, remove_bias, PLearn::TVec< T >::resize(), PLearn::SourceKernel::setDataForKernelMatrix(), PLearn::SourceKernel::source_kernel, PLearn::sum(), total_average, and total_average_unbiased.
{ bool there_was_data_and_it_changed = data && !(data->looksTheSameAs(the_data)); // Set the data for this kernel as well as for the underlying kernel. inherited::setDataForKernelMatrix(the_data); // Check whether we need to recompute the Gram matrix and its average. int n = the_data->length(); if ( data_will_change || average_row.length() != n || there_was_data_and_it_changed) { // Compute the underlying Gram matrix. Mat gram(n, n); source_kernel->computeGramMatrix(gram); // Compute the row (and column) average. average_row.resize(n); average_row.fill(0); if (is_symmetric) { average_col = average_row; } else { average_col.resize(n); average_col.fill(0); } real k_x_x; total_average_unbiased = 0; for (int i = 0; i < n; i++) { if (is_symmetric) { real v; k_x_x = gram(i,i); if (!remove_bias) { average_row[i] += k_x_x; total_average_unbiased -= k_x_x; } for (int j = i + 1; j < n; j++) { v = gram(i,j); average_row[i] += v; average_row[j] += v; } } else { for (int j = 0; j < n; j++) { if (!remove_bias || j != i) { average_row[i] += gram(i,j); average_col[i] += gram(j,i); if (j == i) { total_average_unbiased -= gram(i,j); } } } } } total_average = sum(average_row); if (remove_bias) { // The sum is already unbiased. total_average_unbiased = total_average; } else { // At this point, 'total_average_unbiased' = - \sum K(x_i,x_i). total_average_unbiased += total_average; } real n_terms_in_sum; // The number of terms summed in average_row. if (remove_bias) { // The diagonal terms were not added. n_terms_in_sum = real(n - 1); } else { n_terms_in_sum = real(n); } total_average /= real(n * n_terms_in_sum); total_average_unbiased /= real(n * (n-1)); average_row /= n_terms_in_sum; if (!is_symmetric) { average_col /= n_terms_in_sum; } } }
Reimplemented from PLearn::SourceKernel.
Definition at line 138 of file AdditiveNormalizationKernel.h.
Vec PLearn::AdditiveNormalizationKernel::all_k_x [mutable, private] |
Used to store the values of the source kernel.
Definition at line 60 of file AdditiveNormalizationKernel.h.
Referenced by computeAverage(), and makeDeepCopyFromShallowCopy().
Vec PLearn::AdditiveNormalizationKernel::average_col [protected] |
Definition at line 68 of file AdditiveNormalizationKernel.h.
Referenced by declareOptions(), evaluate_i_j(), evaluate_x_i(), evaluate_x_i_again(), makeDeepCopyFromShallowCopy(), and setDataForKernelMatrix().
Vec PLearn::AdditiveNormalizationKernel::average_row [protected] |
Definition at line 69 of file AdditiveNormalizationKernel.h.
Referenced by declareOptions(), evaluate_i_j(), evaluate_i_x(), evaluate_i_x_again(), makeDeepCopyFromShallowCopy(), and setDataForKernelMatrix().
real PLearn::AdditiveNormalizationKernel::avg_evaluate_i_x_again [mutable, protected] |
The last average computed in evaluate_i_x_again().
Definition at line 76 of file AdditiveNormalizationKernel.h.
Referenced by evaluate_i_x_again().
real PLearn::AdditiveNormalizationKernel::avg_evaluate_x_i_again [mutable, protected] |
The last average computed in evaluate_x_i_again().
Definition at line 79 of file AdditiveNormalizationKernel.h.
Referenced by evaluate_x_i_again().
Definition at line 90 of file AdditiveNormalizationKernel.h.
Referenced by declareOptions(), and setDataForKernelMatrix().
Definition at line 91 of file AdditiveNormalizationKernel.h.
Referenced by build_(), and declareOptions().
real PLearn::AdditiveNormalizationKernel::factor [protected] |
A multiplicative factor to scale the result (1 or -1/2).
Definition at line 82 of file AdditiveNormalizationKernel.h.
Referenced by build_(), evaluate(), evaluate_i_j(), evaluate_i_x(), evaluate_i_x_again(), evaluate_x_i(), and evaluate_x_i_again().
Definition at line 92 of file AdditiveNormalizationKernel.h.
Referenced by declareOptions(), evaluate(), and setDataForKernelMatrix().
Definition at line 93 of file AdditiveNormalizationKernel.h.
Referenced by declareOptions(), and evaluate().
Definition at line 70 of file AdditiveNormalizationKernel.h.
Referenced by declareOptions(), evaluate(), evaluate_i_j(), evaluate_i_x(), evaluate_i_x_again(), evaluate_x_i(), evaluate_x_i_again(), and setDataForKernelMatrix().
Definition at line 71 of file AdditiveNormalizationKernel.h.
Referenced by declareOptions(), evaluate(), and setDataForKernelMatrix().