Skip to content

D
dlib
Commit 04991b7d authored by Davis King's avatar Davis King
Browse files
Options

Made this example program use the new find_max_global() instead of grid search 

and BOBYQA.  This greatly simplifies the example.
parent fc6cce9f
Show whitespace changes
Inline Side-by-side
Showing with 73 additions and 184 deletions
examples/model_selection_ex.cpp
View file @ 04991b7d
// The contents of this file are in the public domain. See LICENSE_FOR_EXAMPLE_PROGRAMS.txt // The contents of this file are in the public domain. See LICENSE_FOR_EXAMPLE_PROGRAMS.txt
/* /*
This is an example that shows some reasonable ways you can perform This is an example that shows how you can perform model selection with the
model selection with the dlib C++ Library. dlib C++ Library.
It will create a simple set of data and then show you how to use It will create a simple dataset and show you how to use cross validation and
the cross validation and optimization routines to determine good model global optimization to determine good parameters for the purpose of training
parameters for the purpose of training an svm to classify the sample data. an svm to classify the data.
The data used in this example will be 2 dimensional data and will The data used in this example will be 2 dimensional data and will come from a
come from a distribution where points with a distance less than 10 distribution where points with a distance less than 10 from the origin are
from the origin are labeled +1 and all other points are labeled labeled +1 and all other points are labeled as -1.
as -1.
As an side, you should probably read the svm_ex.cpp and matrix_ex.cpp example As an side, you should probably read the svm_ex.cpp and matrix_ex.cpp example
...@@ -21,80 +20,22 @@ ...@@ -21,80 +20,22 @@
#include <iostream> #include <iostream>
#include <dlib/svm.h> #include <dlib/svm.h>
#include <dlib/global_optimization.h>
using namespace std; using namespace std;
using namespace dlib; using namespace dlib;
// The svm functions use column vectors to contain a lot of the data on which they
// operate. So the first thing we do here is declare a convenient typedef.
// This typedef declares a matrix with 2 rows and 1 column. It will be the int main() try
// object that contains each of our 2 dimensional samples.
typedef matrix<double, 2, 1> sample_type;
// This is a typedef for the type of kernel we are going to use in this example.
// In this case I have selected the radial basis kernel that can operate on our
// 2D sample_type objects
typedef radial_basis_kernel<sample_type> kernel_type;
class cross_validation_objective
{ {
/*! // The svm functions use column vectors to contain a lot of the data on which they
WHAT THIS OBJECT REPRESENTS // operate. So the first thing we do here is declare a convenient typedef.
This object is a simple function object that takes a set of model
parameters and returns a number indicating how "good" they are. It
does this by performing 10 fold cross validation on our dataset
and reporting the accuracy.
See below in main() for how this object gets used.
!*/
public:
cross_validation_objective (
const std::vector<sample_type>& samples_,
const std::vector<double>& labels_
) : samples(samples_), labels(labels_) {}
double operator() (
const matrix<double>& params
) const
{
// Pull out the two SVM model parameters. Note that, in this case,
// I have setup the parameter search to operate in log scale so we have
// to remember to call exp() to put the parameters back into a normal scale.
const double gamma = exp(params(0));
const double nu = exp(params(1));
// Make an SVM trainer and tell it what the parameters are supposed to be. // This typedef declares a matrix with 2 rows and 1 column. It will be the
svm_nu_trainer<kernel_type> trainer; // object that contains each of our 2 dimensional samples.
trainer.set_kernel(kernel_type(gamma)); typedef matrix<double, 2, 1> sample_type;
trainer.set_nu(nu);
// Finally, perform 10-fold cross validation and then print and return the results.
matrix<double> result = cross_validate_trainer(trainer, samples, labels, 10);
cout << "gamma: " << setw(11) << gamma << " nu: " << setw(11) << nu << " cross validation accuracy: " << result;
// Here I'm returning the harmonic mean between the accuracies of each class.
// However, you could do something else. For example, you might care a lot more
// about correctly predicting the +1 class, so you could penalize results that
// didn't obtain a high accuracy on that class. You might do this by using
// something like a weighted version of the F1-score (see http://en.wikipedia.org/wiki/F1_score).
return 2*prod(result)/sum(result);
}
const std::vector<sample_type>& samples;
const std::vector<double>& labels;
};
int main()
{
try
{
// Now we make objects to contain our samples and their respective labels. // Now we make objects to contain our samples and their respective labels.
std::vector<sample_type> samples; std::vector<sample_type> samples;
...@@ -117,18 +58,17 @@ int main() ...@@ -117,18 +58,17 @@ int main()
labels.push_back(+1); labels.push_back(+1);
else else
labels.push_back(-1); labels.push_back(-1);
} }
} }
cout << "Generated " << samples.size() << " points" << endl; cout << "Generated " << samples.size() << " points" << endl;
// Here we normalize all the samples by subtracting their mean and dividing by their standard deviation. // Here we normalize all the samples by subtracting their mean and dividing by their
// This is generally a good idea since it often heads off numerical stability problems and also // standard deviation. This is generally a good idea since it often heads off
// prevents one large feature from smothering others. Doing this doesn't matter much in this example // numerical stability problems and also prevents one large feature from smothering
// so I'm just doing this here so you can see an easy way to accomplish this with // others. Doing this doesn't matter much in this example so I'm just doing this here
// the library. // so you can see an easy way to accomplish this with the library.
vector_normalizer<sample_type> normalizer; vector_normalizer<sample_type> normalizer;
// let the normalizer learn the mean and standard deviation of the samples // let the normalizer learn the mean and standard deviation of the samples
normalizer.train(samples); normalizer.train(samples);
...@@ -137,129 +77,78 @@ int main() ...@@ -137,129 +77,78 @@ int main()
samples[i] = normalizer(samples[i]); samples[i] = normalizer(samples[i]);
// Now that we have some data we want to train on it. However, there are two parameters to the // Now that we have some data we want to train on it. We are going to train a
// training. These are the nu and gamma parameters. Our choice for these parameters will // binary SVM with the RBF kernel to classify the data. However, there are two
// influence how good the resulting decision function is. To test how good a particular choice // parameters to the training. These are the nu and gamma parameters. Our choice
// of these parameters is we can use the cross_validate_trainer() function to perform n-fold cross // for these parameters will influence how good the resulting decision function is.
// validation on our training data. However, there is a problem with the way we have sampled // To test how good a particular choice of these parameters is we can use the
// our distribution above. The problem is that there is a definite ordering to the samples. // cross_validate_trainer() function to perform n-fold cross validation on our
// That is, the first half of the samples look like they are from a different distribution // training data. However, there is a problem with the way we have sampled our
// than the second half. This would screw up the cross validation process but we can // distribution above. The problem is that there is a definite ordering to the
// fix it by randomizing the order of the samples with the following function call. // samples. That is, the first half of the samples look like they are from a
// different distribution than the second half. This would screw up the cross
// validation process, but we can fix it by randomizing the order of the samples
// with the following function call.
randomize_samples(samples, labels); randomize_samples(samples, labels);
// The nu parameter has a maximum value that is dependent on the ratio of the +1 to -1
// labels in the training data. This function finds that value. The 0.999 is here because
// the maximum allowable nu is strictly less than the value returned by maximum_nu(). So
// rather than dealing with that below we can just back away from it a little bit here and then
// not worry about it.
const double max_nu = 0.999*maximum_nu(labels);
// The first kind of model selection we will do is a simple grid search. That is, below we just // And now we get to the important bit. Here we define a function,
// generate a fixed grid of points (each point represents one possible setting of the model parameters) // cross_validation_score(), that will do the cross-validation we
// and test each via cross validation. // mentioned and return a number indicating how good a particular setting
// of gamma and nu is.
// This code generates a 4x4 grid of logarithmically spaced points. The result is a matrix auto cross_validation_score = [&](const double gamma, const double nu)
// with 2 rows and 16 columns where each column represents one of our points.
matrix<double> params = cartesian_product(logspace(log10(5.0), log10(1e-5), 4), // gamma parameter
logspace(log10(max_nu), log10(1e-5), 4) // nu parameter
);
// As an aside, if you wanted to do a grid search over points of dimensionality more than two
// you would just nest calls to cartesian_product(). You can also use linspace() to generate
// linearly spaced points if that is more appropriate for the parameters you are working with.
// Next we loop over all the points we generated and check how good each is.
cout << "Doing a grid search" << endl;
matrix<double> best_result(2,1);
best_result = 0;
double best_gamma = 0.1, best_nu;
for (long col = 0; col < params.nc(); ++col)
{ {
// pull out the current set of model parameters // Make a RBF SVM trainer and tell it what the parameters are supposed to be.
const double gamma = params(0, col); typedef radial_basis_kernel<sample_type> kernel_type;
const double nu = params(1, col);
// setup a training object using our current parameters
svm_nu_trainer<kernel_type> trainer; svm_nu_trainer<kernel_type> trainer;
trainer.set_kernel(kernel_type(gamma)); trainer.set_kernel(kernel_type(gamma));
trainer.set_nu(nu); trainer.set_nu(nu);
// Finally, do 10 fold cross validation and then check if the results are the best we have seen so far. // Finally, perform 10-fold cross validation and then print and return the results.
matrix<double> result = cross_validate_trainer(trainer, samples, labels, 10); matrix<double> result = cross_validate_trainer(trainer, samples, labels, 10);
cout << "gamma: " << setw(11) << gamma << " nu: " << setw(11) << nu << " cross validation accuracy: " << result; cout << "gamma: " << setw(11) << gamma << " nu: " << setw(11) << nu << " cross validation accuracy: " << result;
// save the best results // Now return a number indicating how good the parameters are. Bigger is
if (sum(result) > sum(best_result)) // better in this example. Here I'm returning the harmonic mean between the
{ // accuracies of each class. However, you could do something else. For
best_result = result; // example, you might care a lot more about correctly predicting the +1 class,
best_gamma = gamma; // so you could penalize results that didn't obtain a high accuracy on that
best_nu = nu; // class. You might do this by using something like a weighted version of the
} // F1-score (see http://en.wikipedia.org/wiki/F1_score).
} return 2*prod(result)/sum(result);
};
cout << "\n best result of grid search: " << sum(best_result) << endl;
cout << " best gamma: " << best_gamma << " best nu: " << best_nu << endl;
// Grid search is a very simple brute force method. Below we try out the BOBYQA algorithm.
// It is a routine that performs optimization of a function in the absence of derivatives.
cout << "\n\n Try the BOBYQA algorithm" << endl;
// We need to supply a starting point for the optimization. Here we are using the best
// result of the grid search. Generally, you want to try and give a reasonable starting
// point due to the possibility of the optimization getting stuck in a local maxima.
params.set_size(2,1);
params = best_gamma, // initial gamma
best_nu; // initial nu
// We also need to supply lower and upper bounds for the search.
matrix<double> lower_bound(2,1), upper_bound(2,1);
lower_bound = 1e-7, // smallest allowed gamma
1e-7; // smallest allowed nu
upper_bound = 100, // largest allowed gamma
max_nu; // largest allowed nu
// The nu parameter has a maximum value that is dependent on the ratio of the +1 to -1
// labels in the training data. This function finds that value. The 0.999 is here
// because the maximum allowable nu is strictly less than the value returned by
// maximum_nu(). So shrinking the limit a little will prevent us from hitting it.
const double max_nu = 0.999*maximum_nu(labels);
// For the gamma and nu SVM parameters it is generally a good idea to search
// in log space. So I'm just converting them into log space here before
// we start the optimization.
params = log(params);
lower_bound = log(lower_bound);
upper_bound = log(upper_bound);
// Finally, ask BOBYQA to look for the best set of parameters. Note that we are using the // And finally, we call this global optimizer that will search for the best parameters.
// cross validation function object defined at the top of the file. // It will call cross_validation_score() 50 times with different settings and return
double best_score = find_max_bobyqa( // the best parameter setting it finds. find_max_global() uses a global optimization
cross_validation_objective(samples, labels), // Function to maximize // method based on a combination of non-parametric global function modeling and
params, // starting point // quadratic trust region modeling to efficiently find a global maximizer. It usually
params.size()*2 + 1, // See BOBYQA docs, generally size*2+1 is a good setting for this // does a good job with a relatively small number of calls to cross_validation_score().
lower_bound, // lower bound // In this example, you should observe that it finds settings that give perfect binary
upper_bound, // upper bound // classification on the data.
min(upper_bound-lower_bound)/10, // search radius auto result = find_max_global(cross_validation_score,
0.01, // desired accuracy {1e-5, 1e-5}, // lower bound constraints on gamma and nu, respectively
100 // max number of allowable calls to cross_validation_objective() {100, max_nu}, // upper bound constraints on gamma and nu, respectively
); max_function_calls(50));
// Don't forget to convert back from log scale to normal scale double best_gamma = result.x(0);
params = exp(params); double best_nu = result.x(1);
cout << " best result of BOBYQA: " << best_score << endl; cout << " best cross-validation score: " << result.y << endl;
cout << " best gamma: " << params(0) << " best nu: " << params(1) << endl; cout << " best gamma: " << best_gamma << " best nu: " << best_nu << endl;
// Also note that the find_max_bobyqa() function only works for optimization problems
// with 2 variables or more. If you only have a single variable then you should use
// the find_max_single_variable() function.
} }
catch (exception& e) catch (exception& e)
{ {
cout << e.what() << endl; cout << e.what() << endl;
}
} }
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