Deprecated API

Contents

• Deprecated Classes
• Deprecated Methods
• Deprecated Constructors
• Deprecated Enum Constants

Deprecated Classes

Class and Description
org.deeplearning4j.earlystopping.scorecalc.DataSetLossCalculatorCG Use DataSetLossCalculator instead for both MultiLayerNetwork and ComputationGraph
org.deeplearning4j.nn.conf.layers.GravesBidirectionalLSTM use Bidirectional instead. With the Bidirectional layer wrapper you can make any recurrent layer bidirectional, in particular GravesLSTM. Note that this layer adds the output of both directions, which translates into "ADD" mode in Bidirectional. Usage: .layer(new Bidirectional(Bidirectional.Mode.ADD, new GravesLSTM.Builder()....build()))

Deprecated Methods

Method and Description
org.deeplearning4j.nn.conf.layers.ActivationLayer.Builder.activation(String) Use ActivationLayer.Builder.activation(Activation) or @activation(IActivation)
org.deeplearning4j.eval.RegressionEvaluation.averagecorrelationR2() Use RegressionEvaluation.averagePearsonCorrelation() instead. For the R2 score use RegressionEvaluation.averageRSquared().
org.deeplearning4j.eval.RegressionEvaluation.correlationR2(int) Use RegressionEvaluation.pearsonCorrelation(int) instead. For the R2 score use RegressionEvaluation.rSquared(int).
org.deeplearning4j.optimize.solvers.accumulation.EncodingHandler.decodeUpdates(INDArray)
org.deeplearning4j.nn.conf.MultiLayerConfiguration.Builder.inferenceWorkspaceMode(WorkspaceMode) Use NeuralNetConfiguration.Builder.inferenceWorkspaceMode(WorkspaceMode)
org.deeplearning4j.nn.api.Model.initParams() As of 0.7.3 - Feb 2017. Not used; neural network params are initialized by the parameter initializaters. Furthermore, most implementations are unsupported or no-op.
org.deeplearning4j.nn.multilayer.MultiLayerNetwork.pretrain(INDArray) use MultiLayerNetwork.pretrain(DataSetIterator) or MultiLayerNetwork.pretrainLayer(int, DataSetIterator) or MultiLayerNetwork.pretrainLayer(int, INDArray). Pretraining each layer in a row on a single minibatch (as per this method) instead of N epochs per layer is not advisable.
org.deeplearning4j.nn.conf.NeuralNetConfiguration.Builder.stepFunction(StepFunction)
org.deeplearning4j.nn.conf.MultiLayerConfiguration.Builder.trainingWorkspaceMode(WorkspaceMode) Use NeuralNetConfiguration.Builder.trainingWorkspaceMode(WorkspaceMode)
org.deeplearning4j.nn.api.Layer.transpose()
org.deeplearning4j.nn.conf.NeuralNetConfiguration.Builder.updater(Updater) Use NeuralNetConfiguration.Builder.updater(IUpdater)
org.deeplearning4j.nn.conf.layers.BaseLayer.Builder.updater(Updater)
org.deeplearning4j.nn.transferlearning.FineTuneConfiguration.Builder.updater(Updater)
org.deeplearning4j.nn.api.Model.validateInput() As of 0.7.3 - Feb 2017. No longer used, most implementations are unsupported or no-op.

Deprecated Constructors

Constructor and Description
org.deeplearning4j.nn.conf.layers.variational.BernoulliReconstructionDistribution(String) Use BernoulliReconstructionDistribution.BernoulliReconstructionDistribution(Activation)
org.deeplearning4j.nn.conf.layers.variational.ExponentialReconstructionDistribution(String) Use ExponentialReconstructionDistribution.ExponentialReconstructionDistribution(Activation)

Deprecated Enum Constants

Enum Constant and Description
org.deeplearning4j.nn.conf.Updater.CUSTOM
org.deeplearning4j.nn.api.OptimizationAlgorithm.HESSIAN_FREE