gov.sandia.cognition.learning.function.scalar

Class LocallyWeightedKernelScalarFunction<InputType>

java.lang.Object

gov.sandia.cognition.util.AbstractCloneableSerializable

gov.sandia.cognition.learning.function.kernel.DefaultKernelContainer<InputType>

gov.sandia.cognition.learning.function.scalar.KernelScalarFunction<InputType>

gov.sandia.cognition.learning.function.scalar.LocallyWeightedKernelScalarFunction<InputType>

Type Parameters:

InputType - Input type to the Kernel

All Implemented Interfaces:

Evaluator<InputType,java.lang.Double>, KernelContainer<InputType>, Regressor<InputType>, ScalarFunction<InputType>, CloneableSerializable, java.io.Serializable, java.lang.Cloneable

public class LocallyWeightedKernelScalarFunction<InputType>
extends KernelScalarFunction<InputType>

The LocallyWeightedKernelScalarFunction class implements a scalar function that uses kernels and does local weighting on them to get the result value. It also includes the ability to bias the function towards a constant when the kernel values are very small.

Since:

2.0

Author:

Justin Basilico

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
protected double	constantValue The constant value is what the constant weight biases the function toward when there is near zero weight.
protected double	constantWeight The constant weight is used as a weight for the constant value that is added to the result to bias the function to the constant value.
static double	DEFAULT_CONSTANT_VALUE The default constant value is 0.0.
static double	DEFAULT_CONSTANT_WEIGHT The default constant weight is 0.0.

Fields inherited from class gov.sandia.cognition.learning.function.scalar.KernelScalarFunction

bias, DEFAULT_BIAS, examples

Fields inherited from class gov.sandia.cognition.learning.function.kernel.DefaultKernelContainer

kernel

Constructor Summary

Constructors

Constructor and Description
LocallyWeightedKernelScalarFunction() Creates a new instance of LocallyWeightedKernelScalarFunction.
LocallyWeightedKernelScalarFunction(Kernel<? super InputType> kernel) Creates a new instance of LocallyWeightedKernelScalarFunction with the given kernel.
LocallyWeightedKernelScalarFunction(Kernel<? super InputType> kernel, java.util.Collection<? extends WeightedValue<? extends InputType>> examples) Creates a new instance of LocallyWeightedKernelScalarFunction with the given kernel and weighted examples.
LocallyWeightedKernelScalarFunction(Kernel<? super InputType> kernel, java.util.Collection<? extends WeightedValue<? extends InputType>> examples, double bias) Creates a new instance of LocallyWeightedKernelScalarFunction with the given kernel, weighted examples, and bias.
LocallyWeightedKernelScalarFunction(Kernel<? super InputType> kernel, java.util.Collection<? extends WeightedValue<? extends InputType>> examples, double bias, double constantWeight, double constantValue) Creates a new instance of LocallyWeightedKernelScalarFunction with the given kernel, weighted examples, and bias.
LocallyWeightedKernelScalarFunction(LocallyWeightedKernelScalarFunction<InputType> other) Creates a new instance copy of LocallyWeightedKernelScalarFunction.

Method Summary

Modifier and Type	Method and Description
java.lang.Double	evaluate(InputType input) Categorizes the given input vector as a double by: (sum w_i * k(input, x_i)) / (sum k(input, x_i))
double	getConstantValue() Gets the constant value.
double	getConstantWeight() Gets the constant weight.
void	setConstantValue(double constantValue) Sets the constant value.
void	setConstantWeight(double constantWeight) Sets the constant weight.

Methods inherited from class gov.sandia.cognition.learning.function.scalar.KernelScalarFunction

evaluateAsDouble, getBias, getExamples, setBias, setExamples

Methods inherited from class gov.sandia.cognition.learning.function.kernel.DefaultKernelContainer

clone, getKernel, setKernel

Methods inherited from class java.lang.Object

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

Field Detail

DEFAULT_CONSTANT_WEIGHT

public static final double DEFAULT_CONSTANT_WEIGHT

The default constant weight is 0.0.

See Also:

Constant Field Values

DEFAULT_CONSTANT_VALUE

public static final double DEFAULT_CONSTANT_VALUE

The default constant value is 0.0.

See Also:

Constant Field Values

constantWeight

protected double constantWeight

The constant weight is used as a weight for the constant value that is added to the result to bias the function to the constant value.

constantValue

protected double constantValue

The constant value is what the constant weight biases the function toward when there is near zero weight.

Constructor Detail

LocallyWeightedKernelScalarFunction

public LocallyWeightedKernelScalarFunction()

Creates a new instance of LocallyWeightedKernelScalarFunction.

LocallyWeightedKernelScalarFunction

public LocallyWeightedKernelScalarFunction(Kernel<? super InputType> kernel)

Creates a new instance of LocallyWeightedKernelScalarFunction with the given kernel.

Parameters:

kernel - The kernel to use.

LocallyWeightedKernelScalarFunction

public LocallyWeightedKernelScalarFunction(Kernel<? super InputType> kernel,
                                           java.util.Collection<? extends WeightedValue<? extends InputType>> examples)

Creates a new instance of LocallyWeightedKernelScalarFunction with the given kernel and weighted examples.

Parameters:

kernel - The kernel to use.

examples - The weighted examples.

LocallyWeightedKernelScalarFunction

public LocallyWeightedKernelScalarFunction(Kernel<? super InputType> kernel,
                                           java.util.Collection<? extends WeightedValue<? extends InputType>> examples,
                                           double bias)

Creates a new instance of LocallyWeightedKernelScalarFunction with the given kernel, weighted examples, and bias.

Parameters:

kernel - The kernel to use.

examples - The weighted examples.

bias - The bias.

LocallyWeightedKernelScalarFunction

public LocallyWeightedKernelScalarFunction(Kernel<? super InputType> kernel,
                                           java.util.Collection<? extends WeightedValue<? extends InputType>> examples,
                                           double bias,
                                           double constantWeight,
                                           double constantValue)

Creates a new instance of LocallyWeightedKernelScalarFunction with the given kernel, weighted examples, and bias.

Parameters:

kernel - The kernel to use.

examples - The weighted examples.

bias - The bias.

constantWeight - The constant bias weight to use.

constantValue - The constant bias value to use.

LocallyWeightedKernelScalarFunction

public LocallyWeightedKernelScalarFunction(LocallyWeightedKernelScalarFunction<InputType> other)

Creates a new instance copy of LocallyWeightedKernelScalarFunction.

Parameters:

other - The object to copy.

Method Detail

evaluate

public java.lang.Double evaluate(InputType input)

Categorizes the given input vector as a double by: (sum w_i * k(input, x_i)) / (sum k(input, x_i))

Specified by:

evaluate in interface Evaluator<InputType,java.lang.Double>

Specified by:

evaluate in interface ScalarFunction<InputType>

Overrides:

evaluate in class KernelScalarFunction<InputType>

Parameters:

input - The input to categorize.

Returns:

The output categorization as a double where the sign is the categorization.

getConstantWeight

public double getConstantWeight()

Gets the constant weight.

Returns:

The constant weight.

setConstantWeight

public void setConstantWeight(double constantWeight)

Sets the constant weight.

Parameters:

constantWeight - The constant weight.

getConstantValue

public double getConstantValue()

Gets the constant value.

Returns:

The constant value.

setConstantValue

public void setConstantValue(double constantValue)

Sets the constant value.

Parameters:

constantValue - The constant value.