org.deeplearning4j.nn.api

Interface Updater

All Superinterfaces:

Cloneable, Serializable

All Known Implementing Classes:

AdaDeltaUpdater, AdaGradUpdater, AdamUpdater, BaseUpdater, MultiLayerUpdater, NesterovsUpdater, NoOpUpdater, RmsPropUpdater, SgdUpdater

public interface Updater
extends Serializable, Cloneable

Update the model

Author:

Adam Gibson

Method Summary

Modifier and Type	Method and Description
Updater	clone()
UpdaterAggregator	getAggregator(boolean addThis) Given this updater, get an UpdaterAggregator that can be used to combine the current state (if any) of this updater with the state of other updaters.
void	update(Layer layer, Gradient gradient, int iteration, int miniBatchSize) Updater: updates the model

Method Detail

update

void update(Layer layer,
            Gradient gradient,
            int iteration,
            int miniBatchSize)

Updater: updates the model

Parameters:

layer -

gradient -

iteration -

getAggregator

UpdaterAggregator getAggregator(boolean addThis)

Given this updater, get an UpdaterAggregator that can be used to combine the current state (if any) of this updater with the state of other updaters. Typically used only in distributed learning scenarios (where each copy of the network has an updater that needs to be combined)

Parameters:

addThis - whether to add the Updater to the UpdaterAggregator, or just return an empy aggregator

clone

Updater clone()