bool fImgSvm::kmeans(SGMatrix<float64_t> &data , CDenseFeatures<float64_t>* ¢ers ,int32_t num_features)
{
init_shogun(&print_message);
int32_t num_clusters= mwordnum ;
int32_t dim_features=SIFTN;
float64_t cluster_std_dev=2.0;
/* build random cluster centers */
SGMatrix<float64_t> cluster_centers(dim_features, num_clusters);
SGVector<float64_t>::random_vector(cluster_centers.matrix, dim_features*num_clusters,
0, 20.0);
//SGMatrix<float64_t>::display_matrix(cluster_centers.matrix, cluster_centers.num_rows,
// cluster_centers.num_cols, "cluster centers");
/* create features, SG_REF to avoid deletion */
CDenseFeatures<float64_t>* features=new CDenseFeatures<float64_t> ();
features->set_feature_matrix(data);
SG_REF(features);
/* create labels for cluster centers */
CMulticlassLabels* labels=new CMulticlassLabels(num_features);
for (index_t i=0; i<num_features; ++i)
labels->set_label(i, 0);
/* create distance */
CEuclideanDistance* distance=new CEuclideanDistance(features, features);
/* create distance machine */
CKMeans* clustering=new CKMeans(num_clusters, distance);
clustering->train(features);
/* build clusters */
// CMulticlassLabels* result=CMulticlassLabels::obtain_from_generic(clustering->apply());
// for (index_t i=0; i<result->get_num_labels(); ++i)
// SG_SPRINT("cluster index of vector %i: %f\n", i, result->get_label(i));
/* print cluster centers */
centers = (CDenseFeatures<float64_t>*)distance->get_lhs();
SGMatrix<float64_t> centers_matrix=centers->get_feature_matrix();
//SG_UNREF(result);
SG_UNREF(centers);
SG_UNREF(clustering);
SG_UNREF(labels);
SG_UNREF(features);
exit_shogun();
}
static std::vector<Datatype, Allocator> ComputeClusterCenters(const std::vector<Datatype, Allocator>& data, const std::vector<u_int32_t>& cluster_labels, std::function<Datatype(const std::vector<Datatype, Allocator>&)>& average_fn, const u_int32_t num_clusters)
{
assert(data.size() == cluster_labels.size());
// Separate the datapoints into their clusters
std::vector<std::vector<Datatype, Allocator>> clustered_data(num_clusters);
for (size_t idx = 0; idx < data.size(); idx++)
{
const Datatype& datapoint = data[idx];
const u_int32_t label = cluster_labels[idx];
clustered_data[label].push_back(datapoint);
}
// Compute the center of each cluster
std::vector<Datatype, Allocator> cluster_centers(num_clusters);
for (u_int32_t cluster = 0; cluster < num_clusters; cluster++)
{
const std::vector<Datatype, Allocator>& cluster_data = clustered_data[cluster];
cluster_centers[cluster] = average_fn(cluster_data);
}
return cluster_centers;
}
