org.deeplearning4j.nn.transferlearning

Class TransferLearning.GraphBuilder

java.lang.Object

org.deeplearning4j.nn.transferlearning.TransferLearning.GraphBuilder

Enclosing class:

TransferLearning

public static class TransferLearning.GraphBuilder
extends java.lang.Object

Constructor Summary

Constructors

Constructor and Description
GraphBuilder(ComputationGraph origGraph) Computation Graph to tweak for transfer learning

Method Summary

Modifier and Type	Method and Description
protected TransferLearning.GraphBuilder	addInputs(java.lang.String... inputNames)
TransferLearning.GraphBuilder	addLayer(java.lang.String layerName, Layer layer, InputPreProcessor preProcessor, java.lang.String... layerInputs) Add a layer with a specified preprocessor
TransferLearning.GraphBuilder	addLayer(java.lang.String layerName, Layer layer, java.lang.String... layerInputs) Add a layer of the specified configuration to the computation graph
TransferLearning.GraphBuilder	addVertex(java.lang.String vertexName, GraphVertex vertex, java.lang.String... vertexInputs) Add a vertex of the given configuration to the computation graph
ComputationGraph	build() Returns a computation graph build to specifications.
TransferLearning.GraphBuilder	fineTuneConfiguration(FineTuneConfiguration fineTuneConfiguration) Set parameters to selectively override existing learning parameters Usage eg.
TransferLearning.GraphBuilder	nOutReplace(java.lang.String layerName, int nOut, Distribution dist) Modify the architecture of a vertex layer by changing nOut Note this will also affect the vertex layer that follows the layer specified, unless it is the output layer Currently does not support modifying nOut of layers that feed into non-layer vertices like merge, subset etc To modify nOut for such vertices use remove vertex, followed by add vertex Can specify different weight init schemes for the specified layer and the layer that follows it.
TransferLearning.GraphBuilder	nOutReplace(java.lang.String layerName, int nOut, Distribution dist, Distribution distNext) Modified nOut of specified layer.
TransferLearning.GraphBuilder	nOutReplace(java.lang.String layerName, int nOut, Distribution dist, WeightInit scheme)
TransferLearning.GraphBuilder	nOutReplace(java.lang.String layerName, int nOut, WeightInit scheme) Modify the architecture of a vertex layer by changing nOut Note this will also affect the vertex layer that follows the layer specified, unless it is the output layer Currently does not support modifying nOut of layers that feed into non-layer vertices like merge, subset etc To modify nOut for such vertices use remove vertex, followed by add vertex Can specify different weight init schemes for the specified layer and the layer that follows it.
TransferLearning.GraphBuilder	nOutReplace(java.lang.String layerName, int nOut, WeightInit scheme, Distribution dist)
TransferLearning.GraphBuilder	nOutReplace(java.lang.String layerName, int nOut, WeightInit scheme, WeightInit schemeNext)
TransferLearning.GraphBuilder	removeVertexAndConnections(java.lang.String vertexName) Remove specified vertex and it's connections from the computation graph
TransferLearning.GraphBuilder	removeVertexKeepConnections(java.lang.String outputName) Remove the specified vertex from the computation graph but keep it's connections.
TransferLearning.GraphBuilder	setFeatureExtractor(java.lang.String... layerName) Specify a layer vertex to set as a "feature extractor" The specified layer vertex and the layers on the path from an input vertex to it it will be "frozen" with parameters staying constant
TransferLearning.GraphBuilder	setOutputs(java.lang.String... outputNames) Set outputs to the computation graph, will add to ones that are existing Also determines the order, like in ComputationGraphConfiguration

Methods inherited from class java.lang.Object

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

Constructor Detail

GraphBuilder

public GraphBuilder(ComputationGraph origGraph)

Computation Graph to tweak for transfer learning

Parameters:

origGraph -

Method Detail

fineTuneConfiguration

public TransferLearning.GraphBuilder fineTuneConfiguration(FineTuneConfiguration fineTuneConfiguration)

Set parameters to selectively override existing learning parameters Usage eg. specify a lower learning rate. This will get applied to all layers

Parameters:

fineTuneConfiguration -

Returns:

GraphBuilder

setFeatureExtractor

public TransferLearning.GraphBuilder setFeatureExtractor(java.lang.String... layerName)

Specify a layer vertex to set as a "feature extractor" The specified layer vertex and the layers on the path from an input vertex to it it will be "frozen" with parameters staying constant

Parameters:

layerName -

Returns:

Builder

nOutReplace

public TransferLearning.GraphBuilder nOutReplace(java.lang.String layerName,
                                                 int nOut,
                                                 WeightInit scheme)

Modify the architecture of a vertex layer by changing nOut Note this will also affect the vertex layer that follows the layer specified, unless it is the output layer Currently does not support modifying nOut of layers that feed into non-layer vertices like merge, subset etc To modify nOut for such vertices use remove vertex, followed by add vertex Can specify different weight init schemes for the specified layer and the layer that follows it.

Parameters:

layerName - The name of the layer to change nOut of

nOut - Value of nOut to change to

scheme - Weight init scheme to use for params in layerName and the layers following it

Returns:

GraphBuilder

See Also:

DISTRIBUTION

nOutReplace

public TransferLearning.GraphBuilder nOutReplace(java.lang.String layerName,
                                                 int nOut,
                                                 Distribution dist)

Modify the architecture of a vertex layer by changing nOut Note this will also affect the vertex layer that follows the layer specified, unless it is the output layer Currently does not support modifying nOut of layers that feed into non-layer vertices like merge, subset etc To modify nOut for such vertices use remove vertex, followed by add vertex Can specify different weight init schemes for the specified layer and the layer that follows it.

Parameters:

layerName - The name of the layer to change nOut of

nOut - Value of nOut to change to

dist - Weight distribution scheme to use

Returns:

GraphBuilder

See Also:

DISTRIBUTION

nOutReplace

public TransferLearning.GraphBuilder nOutReplace(java.lang.String layerName,
                                                 int nOut,
                                                 Distribution dist,
                                                 Distribution distNext)

Modified nOut of specified layer. Also affects layers following layerName unless they are output layers

Parameters:

layerName - The name of the layer to change nOut of

nOut - Value of nOut to change to

dist - Weight distribution scheme to use for layerName

distNext - Weight distribution scheme for layers following layerName

Returns:

GraphBuilder

See Also:

DISTRIBUTION

nOutReplace

public TransferLearning.GraphBuilder nOutReplace(java.lang.String layerName,
                                                 int nOut,
                                                 WeightInit scheme,
                                                 Distribution dist)

nOutReplace

public TransferLearning.GraphBuilder nOutReplace(java.lang.String layerName,
                                                 int nOut,
                                                 Distribution dist,
                                                 WeightInit scheme)

nOutReplace

public TransferLearning.GraphBuilder nOutReplace(java.lang.String layerName,
                                                 int nOut,
                                                 WeightInit scheme,
                                                 WeightInit schemeNext)

removeVertexKeepConnections

public TransferLearning.GraphBuilder removeVertexKeepConnections(java.lang.String outputName)

Remove the specified vertex from the computation graph but keep it's connections. Note the expectation here is to then add back another vertex with the same name or else the graph will be left in an invalid state Possibly with references to vertices that no longer exist

Parameters:

outputName -

Returns:

removeVertexAndConnections

public TransferLearning.GraphBuilder removeVertexAndConnections(java.lang.String vertexName)

Remove specified vertex and it's connections from the computation graph

Parameters:

vertexName -

Returns:

addLayer

public TransferLearning.GraphBuilder addLayer(java.lang.String layerName,
                                              Layer layer,
                                              java.lang.String... layerInputs)

Add a layer of the specified configuration to the computation graph

Parameters:

layerName -

layer -

layerInputs -

Returns:

addLayer

public TransferLearning.GraphBuilder addLayer(java.lang.String layerName,
                                              Layer layer,
                                              InputPreProcessor preProcessor,
                                              java.lang.String... layerInputs)

Add a layer with a specified preprocessor

Parameters:

layerName -

layer -

preProcessor -

layerInputs -

Returns:

addVertex

public TransferLearning.GraphBuilder addVertex(java.lang.String vertexName,
                                               GraphVertex vertex,
                                               java.lang.String... vertexInputs)

Add a vertex of the given configuration to the computation graph

Parameters:

vertexName -

vertex -

vertexInputs -

Returns:

setOutputs

public TransferLearning.GraphBuilder setOutputs(java.lang.String... outputNames)

Set outputs to the computation graph, will add to ones that are existing Also determines the order, like in ComputationGraphConfiguration

Parameters:

outputNames -

Returns:

addInputs

protected TransferLearning.GraphBuilder addInputs(java.lang.String... inputNames)

build

public ComputationGraph build()

Returns a computation graph build to specifications. Init has been internally called. Can be fit directly.

Returns:

Computation graph