程序包 org.apache.flink.ml.common.lossfunc

接口 LossFunc

所有超级接口:

Serializable

所有已知实现类:

BinaryLogisticLoss, HingeLoss, LeastSquareLoss

@Internal
public interface LossFunc
extends Serializable

A loss function is to compute the loss and gradient with the given coefficient and training data.

方法概要

修饰符和类型	方法	说明
void	computeGradient(org.apache.flink.ml.common.feature.LabeledPointWithWeight dataPoint, org.apache.flink.ml.linalg.DenseVector coefficient, org.apache.flink.ml.linalg.DenseVector cumGradient)	Computes the gradient on the given data point and adds the computed gradient to cumGradient.
double	computeLoss(org.apache.flink.ml.common.feature.LabeledPointWithWeight dataPoint, org.apache.flink.ml.linalg.DenseVector coefficient)	Computes the loss on the given data point.

方法详细资料

computeLoss

double computeLoss(org.apache.flink.ml.common.feature.LabeledPointWithWeight dataPoint,
                   org.apache.flink.ml.linalg.DenseVector coefficient)

Computes the loss on the given data point.

参数:

dataPoint - A training data point.

coefficient - The model parameters.

返回:

The loss of the input data.

computeGradient

void computeGradient(org.apache.flink.ml.common.feature.LabeledPointWithWeight dataPoint,
                     org.apache.flink.ml.linalg.DenseVector coefficient,
                     org.apache.flink.ml.linalg.DenseVector cumGradient)

Computes the gradient on the given data point and adds the computed gradient to cumGradient.

参数:

dataPoint - A training data point.

coefficient - The model parameters.

cumGradient - The accumulated gradient.