Java* API Reference for Intel® Data Analytics Acceleration Library 2019 Update 4

List of all members
TrainingModel Class Reference

Class Model object for the training stage of neural network. More...

Class Constructor

TrainingModel ( DaalContext  context)

Constructs model for the training stage of the neural network

Parameters
contextContext to manage the model
TrainingModel ( DaalContext  context,
TrainingModel  other 
)

Constructs model for the training stage of the neural network by copying parameters of another model

Parameters
contextContext to manage the model
otherA model to be used as the source to initialize the parameters of the model

Detailed Description

Member Function Documentation

BackwardLayer getBackwardLayer ( long  index)

Returns the backward stage of a layer with certain index in the network

Parameters
indexIndex of the layer in the network
Returns
Backward stage of a layer with certain index in the network
BackwardLayers getBackwardLayers ( )

Returns list of backward layers

Returns
List of backward layers
ForwardLayer getForwardLayer ( long  index)

Returns the forward stage of a layer with certain index in the network

Parameters
indexIndex of the layer in the network
Returns
Forward stage of a layer with certain index in the network
ForwardLayers getForwardLayers ( )

Returns list of forward layers

Returns
List of forward layers
NextLayersCollection getNextLayers ( )

Returns list of connections between layers

Returns
List of next layers for each layer with corresponding index
PredictionModel getPredictionModel ( Class<?extends Number >  cls)

Returns list of forward layers and their parameters organised in the PredictionModel

Parameters
clsData type to use in intermediate computations for the neural network, Double.class or Float.class
Returns
List of forward layers and their parameters organised in the PredictionModel
NumericTable getWeightsAndBiases ( )

Returns the weights and biases of the forward layer of neural network as numeric table

Returns
The weights and biases numeric table
void initialize ( Class<?extends Number >  cls,
long[]  sampleSize,
TrainingTopology  topology 
)

Initializes neural network

Parameters
clsData type to use in intermediate computations for the neural network, Double.class or Float.class
sampleSizeDimensionality of the training data
topologyCollection of layer descriptors of every inserted layer
void setWeightsAndBiases ( NumericTable  weightsAndBiases)

Sets the weights and biases of the forward layer of neural network as numeric table

Parameters
weightsAndBiasesThe weights and biases numeric table

The documentation for this class was generated from the following file:

For more complete information about compiler optimizations, see our Optimization Notice.