Package org.deeplearning4j.earlystopping.termination

Interface Summary

Interface	Description
EpochTerminationCondition	Interface for termination conditions to be evaluated once per epoch (i.e., once per pass of the full data set), based on a score and epoch number
IterationTerminationCondition	Interface for termination conditions to be evaluated once per iteration (i.e., once per minibatch).

Class Summary

Class	Description
BestScoreEpochTerminationCondition	Created by Sadat Anwar on 3/26/16.
InvalidScoreIterationTerminationCondition	Terminate training at this iteration if score is NaN or Infinite for the last minibatch
MaxEpochsTerminationCondition	Terminate training if the number of epochs exceeds the maximum number of epochs
MaxScoreIterationTerminationCondition	Iteration termination condition for terminating training if the minibatch score exceeds a certain value.
MaxTimeIterationTerminationCondition	Terminate training based on max time.
ScoreImprovementEpochTerminationCondition	Terminate training if best model score does not improve for N epochs