org.nd4j.evaluation.custom

Class CustomEvaluation<T>

java.lang.Object

org.nd4j.evaluation.BaseEvaluation<CustomEvaluation>

org.nd4j.evaluation.custom.CustomEvaluation<T>

All Implemented Interfaces:

Serializable, IEvaluation<CustomEvaluation>

public class CustomEvaluation<T>
extends BaseEvaluation<CustomEvaluation>

A evaluation using lambdas to calculate the score. Uses 3 lambdas:
EvaluationLambda: takes in the labels, predictions, mask, and metadata and returns a value of type T
MergeLambda: takes in two lists of Ts, returns one. Used in merging for distributed training
ResultLambda (in Metric): takes a list of Ts, returns a double value

The EvaluationLambda will be called on each batch, and the results will be stored in a list. MergeLambda merges two of those lists for distributed training (think Spark or Map-Reduce). ResultLambda gets a score from that list.

See Also:

Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	CustomEvaluation.Metric<T> The metric used to get a score for the CustomEvaluation.

Constructor Summary

Constructors

Constructor and Description
CustomEvaluation()

Method Summary

Modifier and Type	Method and Description
void	eval(INDArray labels, INDArray networkPredictions, INDArray maskArray, List<? extends Serializable> recordMetaData)
double	getValue(IMetric metric) Get the value of a given metric for this evaluation.
void	merge(CustomEvaluation other)
static <R> MergeLambda<R>	mergeConcatenate() A MergeLambda that merges by concatenating the two lists
CustomEvaluation<T>	newInstance() Get a new instance of this evaluation, with the same configuration but no data.
void	reset()
String	stats()

Methods inherited from class org.nd4j.evaluation.BaseEvaluation

attempFromLegacyFromJson, eval, eval, eval, evalTimeSeries, evalTimeSeries, fromJson, fromYaml, reshapeAndExtractNotMasked, toJson, toString, toYaml

Methods inherited from class java.lang.Object

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

Constructor Detail

CustomEvaluation

public CustomEvaluation()

Method Detail

mergeConcatenate

public static <R> MergeLambda<R> mergeConcatenate()

A MergeLambda that merges by concatenating the two lists

eval

public void eval(INDArray labels,
                 INDArray networkPredictions,
                 INDArray maskArray,
                 List<? extends Serializable> recordMetaData)

merge

public void merge(CustomEvaluation other)

reset

public void reset()

stats

public String stats()

Returns:

getValue

public double getValue(IMetric metric)

Description copied from interface: IEvaluation

Get the value of a given metric for this evaluation.

newInstance

public CustomEvaluation<T> newInstance()

Description copied from interface: IEvaluation

Get a new instance of this evaluation, with the same configuration but no data.