jsat.lossfunctions

Interface LossFunc

All Superinterfaces:

java.io.Serializable

All Known Subinterfaces:

LossC, LossMC, LossR

All Known Implementing Classes:

AbsoluteLoss, EpsilonInsensitiveLoss, HingeLoss, HuberLoss, LogisticLoss, SoftmaxLoss, SquaredLoss

public interface LossFunc
extends java.io.Serializable

Provides a generic interface for some loss function on some problem that can be described with a single real prediction value and a single real expected value.

A loss function must be non-negative and should be convex.

Method Summary

Modifier and Type	Method and Description
LossFunc	clone()
double	getConjugate(double b, double pred, double y) Computes the result of the conjugate function of this loss.
double	getDeriv(double pred, double y) Computes the first derivative of the loss function.
double	getDeriv2(double pred, double y) Computes the second derivative of the getLoss function.
double	getDeriv2Max() Returns an upper bound on the maximum value of the second derivative.
double	getLoss(double pred, double y) Computes the loss for some problem.
double	lipschitz() If this loss is L-Lipschitz (1/L Lipschitz smooth), this method will return the value of L.

Method Detail

getLoss

double getLoss(double pred,
               double y)

Computes the loss for some problem.

Parameters:

pred - the predicted value in (-Infinity, Infinity)

y - the true value in (-Infinity, Infinity)

Returns:

the loss in [0, Inf)

getDeriv

double getDeriv(double pred,
                double y)

Computes the first derivative of the loss function.

Parameters:

pred - the predicted score in (-Infinity, Infinity)

y - the true value in (-Infinity, Infinity)

Returns:

the first derivative of the getLoss

getDeriv2

double getDeriv2(double pred,
                 double y)

Computes the second derivative of the getLoss function.

Parameters:

pred - the predicted score in (-Infinity, Infinity)

y - the true value in (-Infinity, Infinity)

Returns:

the second derivative of the getLoss function

getConjugate

double getConjugate(double b,
                    double pred,
                    double y)

Computes the result of the conjugate function of this loss. This function is generally optional, and should return Double.NaN if not properly implemented. Many optimization algorithms do require a working implementation though.

Parameters:

b - the primary input to the function

pred - the predicted score in (-Infinity, Infinity)

y - the true class label in {-1, 1}

Returns:

the result of the conjugate function of this loss

getDeriv2Max

double getDeriv2Max()

Returns an upper bound on the maximum value of the second derivative. If the second derivative does not exist, Double.NaN is a valid result. It is also possible for 0 and Double.POSITIVE_INFINITY to be valid results, and must be checked for.

Returns:

the max value of getDeriv2(double, double)

lipschitz

double lipschitz()

If this loss is L-Lipschitz (1/L Lipschitz smooth), this method will return the value of L. If it is not L-Lipschitz, a value of 0 will be returned.

Returns:

the L-Lipschitz constant, or 0 if this loss is not L-Lipschitz;

clone

LossFunc clone()