org.deeplearning4j.arbiter.scoring.impl

Class ROCScoreFunction

java.lang.Object

org.deeplearning4j.arbiter.scoring.impl.BaseNetScoreFunction

org.deeplearning4j.arbiter.scoring.impl.ROCScoreFunction

All Implemented Interfaces:

Serializable, ScoreFunction

public class ROCScoreFunction
extends BaseNetScoreFunction

Score function that calculates AUC (area under ROC curve) or AUPRC (area under precision/recall curve) on a test set for a MultiLayerNetwork or ComputationGraph

Author:

Alex Black

See Also:

Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	ROCScoreFunction.Metric Metric to calculate. AUC: Area under ROC curve AUPRC: Area under precision/recall curve
static class	ROCScoreFunction.ROCType Type of ROC evaluation to perform: ROC: use ROC to perform evaluation (single output binary classification) BINARY: use ROCBinary to perform evaluation (multi-output/multi-task binary classification) MULTICLASS: use ROCMultiClass to perform evaluation (1 vs.

Field Summary

Fields

Modifier and Type	Field and Description
protected ROCScoreFunction.Metric	metric
protected ROCScoreFunction.ROCType	type

Constructor Summary

Constructors

Constructor and Description
ROCScoreFunction(@NonNull ROCScoreFunction.ROCType type, @NonNull ROCScoreFunction.Metric metric)

Method Summary

Modifier and Type	Method and Description
boolean	minimize()
double	score(org.deeplearning4j.nn.graph.ComputationGraph graph, org.nd4j.linalg.dataset.api.iterator.DataSetIterator iterator)
double	score(org.deeplearning4j.nn.graph.ComputationGraph net, org.nd4j.linalg.dataset.api.iterator.MultiDataSetIterator iterator)
double	score(org.deeplearning4j.nn.multilayer.MultiLayerNetwork net, org.nd4j.linalg.dataset.api.iterator.DataSetIterator iterator)
double	score(org.deeplearning4j.nn.multilayer.MultiLayerNetwork net, org.nd4j.linalg.dataset.api.iterator.MultiDataSetIterator iterator)
String	toString()

Methods inherited from class org.deeplearning4j.arbiter.scoring.impl.BaseNetScoreFunction

getSupportedDataTypes, getSupportedModelTypes, score, score, score

Methods inherited from class java.lang.Object

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

Field Detail

type

protected ROCScoreFunction.ROCType type

metric

protected ROCScoreFunction.Metric metric

Constructor Detail

ROCScoreFunction

public ROCScoreFunction(@NonNull
                        @NonNull ROCScoreFunction.ROCType type,
                        @NonNull
                        @NonNull ROCScoreFunction.Metric metric)

Parameters:

type - ROC type to use for evaluation

metric - Evaluation metric to calculate

Method Detail

toString

public String toString()

Overrides:

toString in class Object

score

public double score(org.deeplearning4j.nn.multilayer.MultiLayerNetwork net,
                    org.nd4j.linalg.dataset.api.iterator.DataSetIterator iterator)

Specified by:

score in class BaseNetScoreFunction

score

public double score(org.deeplearning4j.nn.multilayer.MultiLayerNetwork net,
                    org.nd4j.linalg.dataset.api.iterator.MultiDataSetIterator iterator)

Specified by:

score in class BaseNetScoreFunction

score

public double score(org.deeplearning4j.nn.graph.ComputationGraph graph,
                    org.nd4j.linalg.dataset.api.iterator.DataSetIterator iterator)

Specified by:

score in class BaseNetScoreFunction

score

public double score(org.deeplearning4j.nn.graph.ComputationGraph net,
                    org.nd4j.linalg.dataset.api.iterator.MultiDataSetIterator iterator)

Specified by:

score in class BaseNetScoreFunction

minimize

public boolean minimize()