Class SGD<T>

java.lang.Object

org.episteme.core.mathematics.ml.neural.optimizers.SGD<T>

All Implemented Interfaces:

Optimizer<T>

public class SGD<T>
extends Object
implements Optimizer<T>

Stochastic Gradient Descent optimizer.

Constructor Summary

Constructors

Constructor	Description
SGD(double lr)

Method Summary

Modifier and Type	Method	Description
void	step(Map<String, GraphNode<T>> parameters)	Updates the parameters based on their internal gradients.

Methods inherited from class Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface Optimizer

zeroGrad

Constructor Details

SGD

public SGD(double lr)

Method Details

step

public void step(Map<String, GraphNode<T>> parameters)

Description copied from interface: Optimizer

Updates the parameters based on their internal gradients.

Specified by:

step in interface Optimizer<T>

Parameters:

parameters - valid map of parameters to update.