public class GradientDescent extends java.lang.Object implements java.io.Serializable, Disposable
Constructor and Description |
---|
GradientDescent(DataRandomizer dataRandomizer,
SigmoidFunction sigmoidFunction,
CostFunctionCalculator costFunctionCalculator) |
Modifier and Type | Method and Description |
---|---|
void |
dispose()
frees resources associated to this object
|
ProposedUpdate |
performStochasticGradientDescent(java.util.List<DataPoint> trainingData,
java.util.List<Prototype> prototypes,
OmegaMatrix omegaMatrix,
double alphaW,
double alphaO)
performs the stochastic gradient descent on the given data points and
will result in a proposed update which will either be rejected or
accepted and subsequently used in the next epoch
|
public GradientDescent(DataRandomizer dataRandomizer, SigmoidFunction sigmoidFunction, CostFunctionCalculator costFunctionCalculator)
public ProposedUpdate performStochasticGradientDescent(java.util.List<DataPoint> trainingData, java.util.List<Prototype> prototypes, OmegaMatrix omegaMatrix, double alphaW, double alphaO) throws java.lang.InterruptedException, java.util.concurrent.ExecutionException
trainingData
- all training data, however commonly only a subpopulation will
actually be used in this batch learning stepprototypes
- the current prototypesomegaMatrix
- the current mapping rulealphaW
- the prototype learning rate to employalphaO
- the omega learning rate to employjava.lang.InterruptedException
java.util.concurrent.ExecutionException
public void dispose() throws java.lang.InterruptedException
Disposable
dispose
in interface Disposable
java.lang.InterruptedException