Python* API Reference for Intel® Data Analytics Acceleration Library 2020 Update 1

Public Member Functions | Static Public Attributes | List of all members
Batch Class Reference

Computes Stochastic gradient descent in the batch processing mode. More...

Public Member Functions

def __init__
 
def getMethod
 
def getInput
 
def getParameter
 
def createResult
 
def clone
 
def create
 
def compute
 
- Public Member Functions inherited from Batch
def getInput
 
def getParameter
 
def getResult
 
def createResult
 
def clone
 
def compute
 
- Public Member Functions inherited from AlgorithmImpl
def computeNoThrow
 
def compute
 
def checkComputeParams
 
def checkResult
 
def setupCompute
 
def resetCompute
 
def enableResetOnCompute
 
def hostApp
 
def setHostApp
 
- Public Member Functions inherited from Algorithm
def checkComputeParams
 
def getBaseParameter
 
- Public Member Functions inherited from AlgorithmIfaceImpl
def enableChecks
 
def isChecksEnabled
 
- Public Member Functions inherited from AlgorithmIface
def checkComputeParams
 
def checkResult
 
def getMethod
 

Static Public Attributes

 input = ...
 
 parameter = ...
 

Detailed Description

Parameters
fptypeData type to use in intermediate computations for the Stochastic gradient descent algorithm, double or float
methodStochastic gradient descent computation method
Enumerations
  • Method Computation methods for Stochastic gradient descent
  • iterative_solver.InputId Identifiers of input objects for Stochastic gradient descent
  • iterative_solver.ResultId Result identifiers for the Stochastic gradient descent
References
Aliases
  • Batch_Float64DefaultDense is an alias of Batch(fptype=float64, method=daal.algorithms.optimization_solver.sgd.defaultDense)
  • Batch_Float64MiniBatch is an alias of Batch(fptype=float64, method=daal.algorithms.optimization_solver.sgd.miniBatch)
  • Batch_Float64Momentum is an alias of Batch(fptype=float64, method=daal.algorithms.optimization_solver.sgd.momentum)
  • Batch_Float32DefaultDense is an alias of Batch(fptype=float32, method=daal.algorithms.optimization_solver.sgd.defaultDense)
  • Batch_Float32MiniBatch is an alias of Batch(fptype=float32, method=daal.algorithms.optimization_solver.sgd.miniBatch)
  • Batch_Float32Momentum is an alias of Batch(fptype=float32, method=daal.algorithms.optimization_solver.sgd.momentum)

Constructor & Destructor Documentation

def __init__ (   self,
  args 
)

Variant 1

Constructs the SGD algorithm with the input objective function

Parameters
objectiveFunctionObjective function that can be represented as a sum of functions

Variant 2

Constructs the SGD algorithm with the input objective function

Parameters
objectiveFunctionObjective function that can be represented as a sum of functions

Variant 3

Constructs a Stochastic gradient descent algorithm by copying input objects of another Stochastic gradient descent algorithm

Parameters
otherAn algorithm to be used as the source to initialize the input objects and parameters of the algorithm

Member Function Documentation

def clone (   self)

Returns a pointer to the newly allocated Stochastic gradient descent algorithm with a copy of input objects of this Stochastic gradient descent algorithm

Returns
Pointer to the newly allocated algorithm
def compute (   self)

Invokes computations

def create ( )

Creates the instance of the class

Returns
New instance of the class
def createResult (   self)

Creates user-allocated memory to store results of the iterative solver algorithm

Returns
Status of computations
def getInput (   self)

Get input objects for the iterative solver algorithm

Returns
Input objects for the iterative solver algorithm
def getMethod (   self)

Returns method of the algorithm

Returns
Method of the algorithm
def getParameter (   self)

Get parameters of the iterative solver algorithm

Returns
Parameters of the iterative solver algorithm

Member Data Documentation

input = ...
static

Input data structure

parameter = ...
static

Parameter data structure


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

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