org.tribuo.common.libsvm

Class LibSVMTrainer<T extends Output<T>>

java.lang.Object

org.tribuo.common.libsvm.LibSVMTrainer<T>

All Implemented Interfaces:

com.oracle.labs.mlrg.olcut.config.Configurable, com.oracle.labs.mlrg.olcut.provenance.Provenancable<TrainerProvenance>, Trainer<T>

Direct Known Subclasses:

LibSVMAnomalyTrainer, LibSVMClassificationTrainer, LibSVMRegressionTrainer

public abstract class LibSVMTrainer<T extends Output<T>>
extends Object
implements Trainer<T>

A trainer that will train using libsvm's Java implementation.

Note the train method is synchronized on LibSVMTrainer.class due to a global RNG in LibSVM. This is insufficient to ensure reproducibility if LibSVM is used directly in the same JVM as Tribuo, but avoids locking on classes Tribuo does not control.

See:

 Chang CC, Lin CJ.
 "LIBSVM: a library for Support Vector Machines"
 ACM transactions on intelligent systems and technology (TIST), 2011.
 

for the nu-svm algorithm:

 Schölkopf B, Smola A, Williamson R, Bartlett P L.
 "New support vector algorithms"
 Neural Computation, 2000, 1207-1245.
 

and for the original algorithm:

 Cortes C, Vapnik V.
 "Support-Vector Networks"
 Machine Learning, 1995.
 

Field Summary

Fields

Modifier and Type	Field and Description
protected libsvm.svm_parameter	parameters The SVM parameters suitable for use by LibSVM.
protected SVMType<T>	svmType The type of SVM algorithm.

Fields inherited from interface org.tribuo.Trainer

DEFAULT_SEED

Constructor Summary

Constructors

Modifier	Constructor and Description
protected	LibSVMTrainer() For olcut.
protected	LibSVMTrainer(SVMParameters<T> parameters, long seed) Constructs a LibSVMTrainer from the parameters.

Method Summary

Modifier and Type	Method and Description
protected abstract LibSVMModel<T>	createModel(ModelProvenance provenance, ImmutableFeatureMap featureIDMap, ImmutableOutputInfo<T> outputIDInfo, List<libsvm.svm_model> models) Construct the appropriate subtype of LibSVMModel for the prediction task.
static <T extends Output<T>> libsvm.svm_node[]	exampleToNodes(Example<T> example, ImmutableFeatureMap featureIDMap, List<libsvm.svm_node> features) Convert the example into an array of svm_node which represents a sparse feature vector.
protected abstract com.oracle.labs.mlrg.olcut.util.Pair<libsvm.svm_node[][],double[][]>	extractData(Dataset<T> data, ImmutableOutputInfo<T> outputInfo, ImmutableFeatureMap featureMap) Extracts the features and Outputs in LibSVM's format.
int	getInvocationCount() The number of times this trainer instance has had it's train method invoked.
TrainerProvenance	getProvenance()
void	postConfig() Used by the OLCUT configuration system, and should not be called by external code.
protected libsvm.svm_parameter	setupParameters(ImmutableOutputInfo<T> info) Constructs the svm_parameter.
String	toString()
LibSVMModel<T>	train(Dataset<T> examples) Trains a predictive model using the examples in the given data set.
LibSVMModel<T>	train(Dataset<T> examples, Map<String,com.oracle.labs.mlrg.olcut.provenance.Provenance> runProvenance) Trains a predictive model using the examples in the given data set.
protected abstract List<libsvm.svm_model>	trainModels(libsvm.svm_parameter curParams, int numFeatures, libsvm.svm_node[][] features, double[][] outputs, SplittableRandom localRNG) Train all the LibSVM instances necessary for this dataset.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Field Detail

parameters

protected libsvm.svm_parameter parameters

The SVM parameters suitable for use by LibSVM.

svmType

@Config(mandatory=true,
        description="Type of SVM algorithm.")
protected SVMType<T extends Output<T>> svmType

The type of SVM algorithm.

Constructor Detail

LibSVMTrainer

protected LibSVMTrainer()

For olcut.

LibSVMTrainer

protected LibSVMTrainer(SVMParameters<T> parameters,
                        long seed)

Constructs a LibSVMTrainer from the parameters.

Parameters:

parameters - The SVM parameters.

Method Detail

postConfig

public void postConfig()

Used by the OLCUT configuration system, and should not be called by external code.

Specified by:

postConfig in interface com.oracle.labs.mlrg.olcut.config.Configurable

toString

public String toString()

Overrides:

toString in class Object

train

public LibSVMModel<T> train(Dataset<T> examples)

Description copied from interface: Trainer

Trains a predictive model using the examples in the given data set.

Specified by:

train in interface Trainer<T extends Output<T>>

Parameters:

examples - the data set containing the examples.

Returns:

a predictive model that can be used to generate predictions for new examples.

train

public LibSVMModel<T> train(Dataset<T> examples,
                            Map<String,com.oracle.labs.mlrg.olcut.provenance.Provenance> runProvenance)

Description copied from interface: Trainer

Trains a predictive model using the examples in the given data set.

Specified by:

train in interface Trainer<T extends Output<T>>

Parameters:

examples - the data set containing the examples.

runProvenance - Training run specific provenance (e.g., fold number).

Returns:

a predictive model that can be used to generate predictions for new examples.

createModel

protected abstract LibSVMModel<T> createModel(ModelProvenance provenance,
                                              ImmutableFeatureMap featureIDMap,
                                              ImmutableOutputInfo<T> outputIDInfo,
                                              List<libsvm.svm_model> models)

Construct the appropriate subtype of LibSVMModel for the prediction task.

Parameters:

provenance - The model provenance.

featureIDMap - The feature id map.

outputIDInfo - The output id info.

models - The svm models.

Returns:

An implementation of LibSVMModel.

trainModels

protected abstract List<libsvm.svm_model> trainModels(libsvm.svm_parameter curParams,
                                                      int numFeatures,
                                                      libsvm.svm_node[][] features,
                                                      double[][] outputs,
                                                      SplittableRandom localRNG)

Train all the LibSVM instances necessary for this dataset.

Parameters:

curParams - The LibSVM parameters.

numFeatures - The number of features in this dataset.

features - The features themselves.

outputs - The outputs.

localRNG - The RNG to use for seeding LibSVM's RNG.

Returns:

A list of LibSVM models.

extractData

protected abstract com.oracle.labs.mlrg.olcut.util.Pair<libsvm.svm_node[][],double[][]> extractData(Dataset<T> data,
                                                                                                    ImmutableOutputInfo<T> outputInfo,
                                                                                                    ImmutableFeatureMap featureMap)

Extracts the features and Outputs in LibSVM's format.

Parameters:

data - The input data.

outputInfo - The output info.

featureMap - The feature info.

Returns:

The features and outputs.

setupParameters

protected libsvm.svm_parameter setupParameters(ImmutableOutputInfo<T> info)

Constructs the svm_parameter. Most of the time this is a no-op, but classification overrides it to incorporate label weights if they exist.

Parameters:

info - The output info.

Returns:

The svm_parameters to use for training.

getInvocationCount

public int getInvocationCount()

Description copied from interface: Trainer

The number of times this trainer instance has had it's train method invoked.

This is used to determine how many times the trainer's RNG has been accessed to ensure replicability in the random number stream.

Specified by:

getInvocationCount in interface Trainer<T extends Output<T>>

Returns:

The number of train invocations.

exampleToNodes

public static <T extends Output<T>> libsvm.svm_node[] exampleToNodes(Example<T> example,
                                                                     ImmutableFeatureMap featureIDMap,
                                                                     List<libsvm.svm_node> features)

Convert the example into an array of svm_node which represents a sparse feature vector.

If there are collisions in the feature ids then the values are summed.

Type Parameters:

T - The type of the ouput.

Parameters:

example - The example to convert.

featureIDMap - The feature id map which holds the indices.

features - A buffer to use.

Returns:

A sparse feature vector.

getProvenance

public TrainerProvenance getProvenance()

Specified by:

getProvenance in interface com.oracle.labs.mlrg.olcut.provenance.Provenancable<TrainerProvenance>