[ VIGRA Homepage | Function Index | Class Index | Namespaces | File List | Main Page ]

details RandomForest< FEATURES, LABELS, SPLITTESTS, ACCTYPE > Class Template Reference VIGRA

Random forest version 3. More...

#include <vigra/random_forest_3/random_forest.hxx>

Public Member Functions

template<typename IDS >
double leaf_ids (FEATURES const &features, IDS &ids, int n_threads=-1, const std::vector< size_t > tree_indices=std::vector< size_t >()) const
 For each data point in features, compute the corresponding leaf ids and return the average number of split comparisons. More...
 
void merge (RandomForest const &other)
 Grow this forest by incorporating the other.
 
size_t num_classes () const
 Return the number of classes.
 
size_t num_features () const
 Return the number of classes.
 
size_t num_nodes () const
 Return the number of nodes.
 
size_t num_trees () const
 Return the number of trees.
 
void predict (FEATURES const &features, LABELS &labels, int n_threads=-1, const std::vector< size_t > &tree_indices=std::vector< size_t >()) const
 Predict the given data and return the average number of split comparisons. More...
 
template<typename PROBS >
void predict_probabilities (FEATURES const &features, PROBS &probs, int n_threads=-1, const std::vector< size_t > &tree_indices=std::vector< size_t >()) const
 Predict the probabilities of the given data and return the average number of split comparisons. More...
 

Public Attributes

Graph graph_
 The graph structure.
 
NodeMap< AccInputType >::type node_responses_
 Contains the responses of each node (for example the most frequent label).
 
RandomForestOptions options_
 The options that were used for training.
 
ProblemSpec< LabelType > problem_spec_
 The specifications.
 
NodeMap< SplitTests >::type split_tests_
 Contains a test for each internal node, that is used to determine whether given data goes to the left or the right child.
 

Detailed Description

template<typename FEATURES, typename LABELS, typename SPLITTESTS = LessEqualSplitTest<typename FEATURES::value_type>, typename ACCTYPE = ArgMaxVectorAcc<double>>
class vigra::rf3::RandomForest< FEATURES, LABELS, SPLITTESTS, ACCTYPE >

Random forest version 3.

vigra::rf3::RandomForest is typicall constructed via the factory function vigra::rf3::random_forest().

Member Function Documentation

void predict ( FEATURES const &  features,
LABELS &  labels,
int  n_threads = -1,
const std::vector< size_t > &  tree_indices = std::vector<size_t>() 
) const

Predict the given data and return the average number of split comparisons.

Note
labels must be a 1-D array with size features.shape(0).
void predict_probabilities ( FEATURES const &  features,
PROBS &  probs,
int  n_threads = -1,
const std::vector< size_t > &  tree_indices = std::vector<size_t>() 
) const

Predict the probabilities of the given data and return the average number of split comparisons.

Note
probs should have the shape (features.shape()[0], num_classes).
double leaf_ids ( FEATURES const &  features,
IDS &  ids,
int  n_threads = -1,
const std::vector< size_t >  tree_indices = std::vector<size_t>() 
) const

For each data point in features, compute the corresponding leaf ids and return the average number of split comparisons.

Note
ids should have the shape (features.shape()[0], num_trees).

The documentation for this class was generated from the following file:

© Ullrich Köthe (ullrich.koethe@iwr.uni-heidelberg.de)
Heidelberg Collaboratory for Image Processing, University of Heidelberg, Germany

html generated using doxygen and Python
vigra 1.11.0 (Fri May 19 2017)