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

List of all members
DistributedStep2Master Class Reference

Computes neural network training in the distributed processing mode on the master node. More...

Class Constructor

Constructs the neural network training algorithm by copying input objects and parameters of another neural network training algorithm

Parameters
contextContext to manage algorithm
otherAn algorithm to be used as the source to initialize the input objects and parameters of the algorithm
DistributedStep2Master ( DaalContext  context,
Class<?extends Number >  cls,
TrainingMethod  method,
Batch  optimizationSolver 
)

Constructs the neural network training algorithm

Parameters
contextContext to manage algorithm
clsData type to use in intermediate computations for the algorithm, Double.class or Float.class
methodComputation method of the algorithm, TrainingMethod
optimizationSolverOptimization solver
DistributedStep2Master ( DaalContext  context,
Batch  optimizationSolver 
)

Constructs neural network model-based training object with Float data type used for intermediate computations and default computation method

Parameters
contextContext to manage the neural network training object
optimizationSolverOptimization solver

Detailed Description

Member Function Documentation

DistributedStep2Master clone ( DaalContext  context)

Returns the newly allocated neural network training algorithm with a copy of input objects and parameters of this neural network training algorithm

Parameters
contextContext to manage algorithm
Returns
The newly allocated algorithm

Runs the neural network training algorithm on master

Returns
Partial results of the neural network training algorithm
TrainingResult finalizeCompute ( )

Computes the results of the neural network training algorithm on master

Returns
Results of the neural network training algorithm
TrainingResult getResult ( )

Registers user-allocated memory to store the results of the neural network training algorithm

Returns
Results of the neural network training algorithm
void initialize ( long[]  dataSize,
TrainingTopology  topology 
)

Initializes the neural network topology of the layers

Parameters
dataSizeSize of the input data for the trainings
topologyTrainingTopology of the layers
void setPartialResult ( DistributedPartialResult  partialResult)

Registers user-allocated memory to store partial results of the neural network training algorithm

Parameters
partialResultStructure to store partial results of the neural network training algorithm

Member Data Documentation

Input data

Computation method for the algorithm

TrainingParameter parameter

Parameters of the algorithm


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

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