Parameters of the LBFGS algorithm.
More...
Constructs the parameters of the LBFGS algorithm
- Parameters
-
context | Context to manage the parameters of the LBFGS algorithm |
Constructs the parameter for the LBFGS algorithm
- Parameters
-
context | Context to manage the LBFGS algorithm |
cObject | Pointer to C++ implementation of the parameter |
Gets the numeric table that represents 32 bit integer indices of terms of the objective function.
- Returns
- The numeric table that represents 32 bit integer indices of terms of the objective function
Gets the numeric table that represents 32 bit integer indices of terms of the objective function.
- Returns
- The numeric table that represents 32 bit integer indices of terms of the objective function
long getCorrectionPairBatchSize |
( |
| ) |
|
Gets the number of observations to compute the sub-sampled Hessian for correction pairs computation
- Returns
- The number of observations to compute the sub-sampled Hessian for correction pairs computation
Gets the number of iterations between the curvature estimates calculations
- Returns
- The number of iterations between the curvature estimates calculations
Gets the memory parameter of the LBFGS algorithm. Which is the maximum number of correction pairs that define the approximation of inverse Hessian matrix
- Returns
- The memory parameter of the LBFGS algorithm
- Deprecated:
- This item will be removed in a future release.
Gets the seed for random generation of 32 bit integer indices of terms in the objective function.
- Returns
- The seed for random generation of 32 bit integer indices of terms in the objective function.
Gets the numeric table that contains values of the step-length sequence
- Returns
- The numeric table that contains values of the step-length sequence
Sets the numeric table of size nIterations x batchSize that represent indices that will be used instead of random values for the stochastic gradient computations. If no indices are provided, the implementation will generate random indices.
- Parameters
-
batchIndices | The numeric table that represents 32 bit integer indices of terms of the objective function |
void setCorrectionPairBatchIndices |
( |
NumericTable |
batchIndices | ) |
|
Sets the numeric table of size (nIterations / L) x correctionPairBatchSize that represent indices that will be used instead of random values for the sub-sampled Hessian matrix computations. If no indices are provided, the implementation will generate random indices.
- Parameters
-
batchIndices | The numeric table that represents 32 bit integer indices of terms of the objective function |
void setCorrectionPairBatchSize |
( |
long |
batchSize | ) |
|
Sets the number of observations to compute the sub-sampled Hessian for correction pairs computation
- Parameters
-
batchSize | The number of observations to compute the sub-sampled Hessian for correction pairs computation |
Sets the engine to be used by the algorithm
- Parameters
-
engine | to be used by the algorithm |
Sets the number of iterations between the curvature estimates calculations
- Parameters
-
L | The number of iterations between the curvature estimates calculations |
Sets the memory parameter of the LBFGS algorithm. Which is the maximum number of correction pairs that define the approximation of inverse Hessian matrix
- Parameters
-
m | The memory parameter of the LBFGS algorithm |
void setSeed |
( |
long |
seed | ) |
|
- Deprecated:
- This item will be removed in a future release.
Sets the seed for random generation of 32 bit integer indices of terms in the objective function.
- Parameters
-
seed | The seed for random generation of 32 bit integer indices of terms in the objective function. |
void setStepLengthSequence |
( |
NumericTable |
stepLengthSequence | ) |
|
Sets the numeric table that contains values of the step-length sequence
- Parameters
-
stepLengthSequence | The numeric table that contains values of the step-length sequence |
The documentation for this class was generated from the following file:
- optimization_solver/lbfgs/Parameter.java