org.apache.ignite.ml.nn.architecture

Class MLPArchitecture

java.lang.Object

org.apache.ignite.ml.nn.architecture.MLPArchitecture

All Implemented Interfaces:

Serializable

public class MLPArchitecture
extends Object
implements Serializable

Class containing information about architecture of MLP.

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
MLPArchitecture(int inputSize) Construct an MLP architecture.

Method Summary

Modifier and Type	Method and Description
MLPArchitecture	add(MLPArchitecture second) Creates config describing network where first goes this config and after goes this method's argument.
int	inputSize() Size of input of MLP.
LayerArchitecture	layerArchitecture(int layer) Get architecture of layer with given index.
int	layersCount() Count of layers in MLP.
int	outputSize() Size of output of MLP.
int	parametersCount() Count of parameters in this MLP architecture.
TransformationLayerArchitecture	transformationLayerArchitecture(int layer) Get architecture of transformation layer (i.e. non-input layer) with given index.
MLPArchitecture	withAddedLayer(int neuronsCnt, boolean hasBias, IgniteDifferentiableDoubleToDoubleFunction f) Constructs new MLP architecture with new layer added on top of all this architecture layers.

Methods inherited from class java.lang.Object

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

Constructor Detail

MLPArchitecture

public MLPArchitecture(int inputSize)

Construct an MLP architecture.

Parameters:

inputSize - Size of input to MLP.

Method Detail

layersCount

public int layersCount()

Count of layers in MLP.

Returns:

Layers count.

inputSize

public int inputSize()

Size of input of MLP.

Returns:

Size of input.

outputSize

public int outputSize()

Size of output of MLP.

Returns:

Size of output.

withAddedLayer

public MLPArchitecture withAddedLayer(int neuronsCnt,
                                      boolean hasBias,
                                      IgniteDifferentiableDoubleToDoubleFunction f)

Constructs new MLP architecture with new layer added on top of all this architecture layers.

Parameters:

neuronsCnt - Count of neurons in new layer.

hasBias - Flag indicating presence of bias in added layer.

f - Activation function of a new layer.

Returns:

New MLP architecture with new layer added on top of all this architecture layers.

layerArchitecture

public LayerArchitecture layerArchitecture(int layer)

Get architecture of layer with given index.

Parameters:

layer - Index of layer to get architecture from.

Returns:

Architecture of layer with given index.

transformationLayerArchitecture

public TransformationLayerArchitecture transformationLayerArchitecture(int layer)

Get architecture of transformation layer (i.e. non-input layer) with given index.

Parameters:

layer - Index of layer to get architecture from.

Returns:

Architecture of transformation layer with given index.

add

public MLPArchitecture add(MLPArchitecture second)

Creates config describing network where first goes this config and after goes this method's argument.

Parameters:

second - Config to add after this config.

Returns:

New combined configuration.

parametersCount

public int parametersCount()

Count of parameters in this MLP architecture.

Returns:

Parameters in this MLP architecture.