Package org.apache.commons.math4.neuralnet

Class FeatureInitializerFactory

java.lang.Object

org.apache.commons.math4.neuralnet.FeatureInitializerFactory

public final class FeatureInitializerFactory
extends Object

Creates functions that will select the initial values of a neuron's features.

Since:

3.3

Method Summary

Modifier and Type	Method	Description
static FeatureInitializer	function(DoubleUnaryOperator f, double init, double inc)	Creates an initializer from a univariate function f(x).
static FeatureInitializer	randomize(org.apache.commons.rng.sampling.distribution.ContinuousUniformSampler random, FeatureInitializer orig)	Adds some amount of random data to the given initializer.
static FeatureInitializer	uniform(org.apache.commons.rng.UniformRandomProvider rng, double min, double max)	Uniform sampling of the given range.

Methods inherited from class java.lang.Object

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

Method Detail

uniform

public static FeatureInitializer uniform(org.apache.commons.rng.UniformRandomProvider rng,
                                         double min,
                                         double max)

Uniform sampling of the given range.

Parameters:

min - Lower bound of the range.

max - Upper bound of the range.

rng - Random number generator used to draw samples from a uniform distribution.

Returns:

an initializer such that the features will be initialized with values within the given range.

Throws:

IllegalArgumentException - if min >= max.

function

public static FeatureInitializer function(DoubleUnaryOperator f,
                                          double init,
                                          double inc)

Creates an initializer from a univariate function f(x). The argument x is set to init at the first call and will be incremented at each call.

Parameters:

f - Function.

init - Initial value.

inc - Increment

Returns:

the initializer.

randomize

public static FeatureInitializer randomize(org.apache.commons.rng.sampling.distribution.ContinuousUniformSampler random,
                                           FeatureInitializer orig)

Adds some amount of random data to the given initializer.

Parameters:

random - Random variable distribution sampler.

orig - Original initializer.

Returns:

an initializer whose value method will return orig.value() + random.sample().