org.apache.ignite.ml.nn.initializers

Class RandomInitializer

java.lang.Object

org.apache.ignite.ml.nn.initializers.RandomInitializer

All Implemented Interfaces:

MLPInitializer

public class RandomInitializer
extends Object
implements MLPInitializer

Class for initialization of MLP parameters with random uniformly distributed numbers from -1 to 1.

Constructor Summary

Constructors

Constructor and Description
RandomInitializer() Constructs RandomInitializer with random seed.
RandomInitializer(long seed) Constructs RandomInitializer with the given seed.
RandomInitializer(Random rnd) Construct RandomInitializer from given RNG.

Method Summary

Modifier and Type	Method and Description
void	initBiases(Vector biases) In-place change values of vector representing vectors.
void	initWeights(Matrix weights) In-place change values of matrix representing weights.

Methods inherited from class java.lang.Object

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

Constructor Detail

RandomInitializer

public RandomInitializer(Random rnd)

Construct RandomInitializer from given RNG.

Parameters:

rnd - RNG.

RandomInitializer

public RandomInitializer(long seed)

Constructs RandomInitializer with the given seed.

Parameters:

seed - Seed.

RandomInitializer

public RandomInitializer()

Constructs RandomInitializer with random seed.

Method Detail

initWeights

public void initWeights(Matrix weights)

In-place change values of matrix representing weights.

Specified by:

initWeights in interface MLPInitializer

Parameters:

weights - Matrix representing weights.

initBiases

public void initBiases(Vector biases)

In-place change values of vector representing vectors.

Specified by:

initBiases in interface MLPInitializer

Parameters:

biases - Vector representing vectors.