jsat.lossfunctions

Class HingeLoss

java.lang.Object

jsat.lossfunctions.HingeLoss

All Implemented Interfaces:

java.io.Serializable, LossC, LossFunc, LossMC

public class HingeLoss
extends java.lang.Object
implements LossMC

The HingeLoss loss function for classification L(x, y) = max(0, 1-y*x) . This also includes the multi-class version of the hinge loss.
This function is only once differentiable.

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
HingeLoss()

Method Summary

Modifier and Type	Method and Description
static CategoricalResults	classify(double score)
HingeLoss	clone()
static double	deriv(double pred, double y) Computes the first derivative of the HingeLoss loss
void	deriv(Vec processed, Vec derivs, int y) Computes the derivatives with respect to each output processed and derivs may be the same object, and will simply have all its values altered if so.
CategoricalResults	getClassification(double score) Given the score value of a data point, this returns the classification results.
CategoricalResults	getClassification(Vec processed) Given the processed predictions, returns the classification results for said predictions.
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 getLoss 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 getLoss for a classification problem.
double	getLoss(Vec processed, int y) Computes the scalar loss for on the given example
double	lipschitz() If this loss is L-Lipschitz (1/L Lipschitz smooth), this method will return the value of L.
static double	loss(double pred, double y) Computes the HingeLoss loss
void	process(Vec pred, Vec processed) Given the vector of raw outputs for each class, transform it into a new vector.

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

HingeLoss

public HingeLoss()

Method Detail

loss

public static double loss(double pred,
                          double y)

Computes the HingeLoss loss

Parameters:

pred - the predicted value

y - the target value

Returns:

the HingeLoss loss

deriv

public static double deriv(double pred,
                           double y)

Computes the first derivative of the HingeLoss loss

Parameters:

pred - the predicted value

y - the target value

Returns:

the first derivative of the HingeLoss loss

classify

public static CategoricalResults classify(double score)

getLoss

public double getLoss(double pred,
                      double y)

Description copied from interface: LossC

Computes the getLoss for a classification problem.

Specified by:

getLoss in interface LossC

Specified by:

getLoss in interface LossFunc

Parameters:

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

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

Returns:

the getLoss in [0, Inf)

getDeriv

public double getDeriv(double pred,
                       double y)

Description copied from interface: LossC

Computes the first derivative of the getLoss function.

Specified by:

getDeriv in interface LossC

Specified by:

getDeriv in interface LossFunc

Parameters:

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

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

Returns:

the first derivative of the getLoss

getDeriv2

public double getDeriv2(double pred,
                        double y)

Description copied from interface: LossC

Computes the second derivative of the getLoss function.

Specified by:

getDeriv2 in interface LossC

Specified by:

getDeriv2 in interface LossFunc

Parameters:

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

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

Returns:

the second derivative of the getLoss function

getDeriv2Max

public double getDeriv2Max()

Description copied from interface: LossFunc

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.

Specified by:

getDeriv2Max in interface LossFunc

Returns:

the max value of LossFunc.getDeriv2(double, double)

clone

public HingeLoss clone()

Specified by:

clone in interface LossC

Specified by:

clone in interface LossFunc

Overrides:

clone in class java.lang.Object

getClassification

public CategoricalResults getClassification(double score)

Description copied from interface: LossC

Given the score value of a data point, this returns the classification results.

Specified by:

getClassification in interface LossC

Parameters:

score - the score for a data point

Returns:

the categorical results with the correct probability values for this loss function.

getLoss

public double getLoss(Vec processed,
                      int y)

Description copied from interface: LossMC

Computes the scalar loss for on the given example

Specified by:

getLoss in interface LossMC

Parameters:

processed - the vector of raw predictions.

y - the true class label in [0, k-1] for k classes

Returns:

the loss in [0, Inf)

process

public void process(Vec pred,
                    Vec processed)

Description copied from interface: LossMC

Given the vector of raw outputs for each class, transform it into a new vector.
processed and derivs may be the same object, and will simply have all its values altered if so.

Specified by:

process in interface LossMC

Parameters:

pred - the vector of raw predictions

processed - the location to store the processed predictions.

deriv

public void deriv(Vec processed,
                  Vec derivs,
                  int y)

Description copied from interface: LossMC

Computes the derivatives with respect to each output
processed and derivs may be the same object, and will simply have all its values altered if so.

Specified by:

deriv in interface LossMC

Parameters:

processed - the processed predictions

derivs - the vector to place the derivative of the loss to.

y - the true class label in [0, k-1] for k classes

getClassification

public CategoricalResults getClassification(Vec processed)

Description copied from interface: LossMC

Given the processed predictions, returns the classification results for said predictions.

Specified by:

getClassification in interface LossMC

Parameters:

processed - the processed score/prediction vector

Returns:

the classification results

getConjugate

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

Description copied from interface: LossFunc

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.

Specified by:

getConjugate in interface LossFunc

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

lipschitz

public double lipschitz()

Description copied from interface: LossFunc

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.

Specified by:

lipschitz in interface LossFunc

Returns:

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