nuroko.module

Class ALayerStack

java.lang.Object

nuroko.module.ALayerStack

All Implemented Interfaces:

Cloneable, IInput, IInputOutput, IModule, IOutput, IParameterised, IThinker

Direct Known Subclasses:

CompoundStack, NeuralNet

public abstract class ALayerStack
extends Object
implements IThinker, IParameterised

Abstract base class representing a stack of WeightLayer components

Author:

Mike

Constructor Summary

Constructors

Constructor and Description
ALayerStack()

Method Summary

Methods

Modifier and Type	Method and Description
abstract ALayerStack	clone() Creates a clone of a module, including a deep copy of any mutable state.
abstract mikera.vectorz.AVector	getData(int i)
mikera.vectorz.AVector	getInput()
abstract AWeightLayer	getLayer(int i)
abstract int	getLayerCount()
List<AWeightLayer>	getLayers()
mikera.vectorz.AVector	getOutput()
int	getParameterLength() Returns the length of the parameter vector for this model
abstract void	train(mikera.vectorz.AVector input, mikera.vectorz.AVector target)
abstract void	trainGradient(mikera.vectorz.AVector input, mikera.vectorz.AVector outputGradient, mikera.vectorz.AVector inputGradient, double factor, boolean skipTopDerivative) Trains with a direct gradient.

Methods inherited from class java.lang.Object

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

Methods inherited from interface nuroko.core.IThinker

think

Methods inherited from interface nuroko.core.IOutput

getOutputLength

Methods inherited from interface nuroko.core.IInput

getInputLength

Methods inherited from interface nuroko.core.IModule

getComponents

Methods inherited from interface nuroko.core.IParameterised

getGradient, getParameters

Constructor Detail

ALayerStack

public ALayerStack()

Method Detail

getLayerCount

public abstract int getLayerCount()

getLayer

public abstract AWeightLayer getLayer(int i)

getData

public abstract mikera.vectorz.AVector getData(int i)

getOutput

public mikera.vectorz.AVector getOutput()

getInput

public mikera.vectorz.AVector getInput()

getLayers

public List<AWeightLayer> getLayers()

clone

public abstract ALayerStack clone()

Description copied from interface: IModule

Creates a clone of a module, including a deep copy of any mutable state.

Specified by:

clone in interface IModule

Specified by:

clone in interface IParameterised

Specified by:

clone in interface IThinker

Overrides:

clone in class Object

Returns:

getParameterLength

public int getParameterLength()

Description copied from interface: IParameterised

Returns the length of the parameter vector for this model

Specified by:

getParameterLength in interface IParameterised

Returns:

train

public abstract void train(mikera.vectorz.AVector input,
         mikera.vectorz.AVector target)

trainGradient

public abstract void trainGradient(mikera.vectorz.AVector input,
                 mikera.vectorz.AVector outputGradient,
                 mikera.vectorz.AVector inputGradient,
                 double factor,
                 boolean skipTopDerivative)

Trains with a direct gradient. Assumes think has been called immediately prior to set any intermediate values.