org.nd4j.linalg.lossfunctions.impl.LossMSE Maven / Gradle / Ivy
package org.nd4j.linalg.lossfunctions.impl;
import lombok.EqualsAndHashCode;
import org.nd4j.linalg.activations.IActivation;
import org.nd4j.linalg.api.ndarray.INDArray;
/**
* Mean Squared Error loss function: L = 1/N sum_i (actual_i - predicted)^2
* See also {@link LossL2} for a mathematically similar loss function (LossL2 does not have division by N, where N is output size)
*
* @author Susan Eraly
*/
@EqualsAndHashCode(callSuper = true)
public class LossMSE extends LossL2 {
public LossMSE() {
}
/**
* Mean Squared Error loss function where each the output is (optionally) weighted/scaled by a fixed scalar value.
* Note that the weights array must be a row vector, of length equal to the labels/output dimension 1 size.
* A weight vector of 1s should give identical results to no weight vector.
*
* @param weights Weights array (row vector). May be null.
*/
public LossMSE(INDArray weights) {
super(weights);
}
@Override
public double computeScore(INDArray labels, INDArray preOutput, IActivation activationFn, INDArray mask, boolean average) {
double score = super.computeScore(labels, preOutput, activationFn, mask, average);
score /= (labels.size(1));
return score;
}
@Override
public INDArray computeScoreArray(INDArray labels, INDArray preOutput, IActivation activationFn, INDArray mask) {
INDArray scoreArr = super.computeScoreArray(labels, preOutput, activationFn, mask);
return scoreArr.divi(labels.size(1));
}
@Override
public INDArray computeGradient(INDArray labels, INDArray preOutput, IActivation activationFn, INDArray mask) {
INDArray gradients = super.computeGradient(labels, preOutput, activationFn, mask);
return gradients.divi(labels.size(1));
}
@Override
public String toString() {
if (weights == null) return "LossMSE()";
return "LossMSE(weights=" + weights + ")";
}
}