API Reference - RecurrentModels

Constructors

RecurrentNeuralNetworkCell() 


RecurrentModels.RecurrentNeuralNetworkCell{inputDimensionSize: number, hiddenDimensionSize: number, learningRate: number, activationFunction: string}: function, WeightContainer, function, function

Parameters:

  • inputDimensionSize: The number of features it takes as inputs.

  • hiddenDimensionSize: The number of features it will produce.

  • learningRate: The speed at which the model learns. Recommended that the value is set between 0 to 1.

  • activationFunction: The activation function to be used for weight activation. [Default: FastTanh]

Returns:

  • Model: The model that is constructed using the set of parameters.

  • WeightContainer: The generated WeightContainer object.

  • reset: The function to reset the hidden state.

  • setHiddenStateTensor: The function to set the hidden state tensor.

GatedRecurrentUnitCell() 


RecurrentModels.GatedRecurrentUnitCell{inputDimensionSize: number, hiddenDimensionSize: number, learningRate: number}: function, WeightContainer, function, function

Parameters:

  • inputDimensionSize: The number of features it takes as inputs.

  • hiddenDimensionSize: The number of features it will produce.

  • learningRate: The speed at which the model learns. Recommended that the value is set between 0 to 1.

Returns:

  • Model: The model that is constructed using the set of parameters.

  • WeightContainer: The generated WeightContainer object.

  • reset: The function to reset the hidden state.

  • setHiddenStateTensor: The function to set the hidden state tensor.

MinimalGatedUnitCell() 


RecurrentModels.MinimalGatedUnitCell{inputDimensionSize: number, hiddenDimensionSize: number, learningRate: number}: function, WeightContainer, function, function

Parameters:

  • inputDimensionSize: The number of features it takes as inputs.

  • hiddenDimensionSize: The number of features it will produce.

  • learningRate: The speed at which the model learns. Recommended that the value is set between 0 to 1.

Returns:

  • Model: The model that is constructed using the set of parameters.

  • WeightContainer: The generated WeightContainer object.

  • reset: The function to reset the hidden state.

  • setHiddenStateTensor: The function to set the hidden state tensor.

LightRecurrentUnitCell() 


RecurrentModels.LightRecurrentUnitCell{inputDimensionSize: number, hiddenDimensionSize: number, learningRate: number}: function, WeightContainer, function, function

Parameters:

  • inputDimensionSize: The number of features it takes as inputs.

  • hiddenDimensionSize: The number of features it will produce.

  • learningRate: The speed at which the model learns. Recommended that the value is set between 0 to 1.

Returns:

  • Model: The model that is constructed using the set of parameters.

  • WeightContainer: The generated WeightContainer object.

  • reset: The function to reset the hidden state.

  • setHiddenStateTensor: The function to set the hidden state tensor.

SimpleRecurrentUnitCell() 


RecurrentModels.SimpleRecurrentUnitCell{inputDimensionSize: number, hiddenDimensionSize: number, learningRate: number}: function, WeightContainer, function, function

Parameters:

  • inputDimensionSize: The number of features it takes as inputs.

  • hiddenDimensionSize: The number of features it will produce.

  • learningRate: The speed at which the model learns. Recommended that the value is set between 0 to 1.

Returns:

  • Model: The model that is constructed using the set of parameters.

  • WeightContainer: The generated WeightContainer object.

  • reset: The function to reset the hidden state.

  • setCellStateTensor: The function to set the cell state tensor.

LongShortTermMemoryCell() 


RecurrentModels.LongShortTermMemoryCell{inputDimensionSize: number, hiddenDimensionSize: number, learningRate: number}: function, WeightContainer, function, function

Parameters:

  • inputDimensionSize: The number of features it takes as inputs.

  • hiddenDimensionSize: The number of features it will produce.

  • learningRate: The speed at which the model learns. Recommended that the value is set between 0 to 1.

Returns:

  • Model: The model that is constructed using the set of parameters.

  • WeightContainer: The generated WeightContainer object.

  • reset: The function to reset the hidden state.

  • setHiddenStateTensor: The function to set the hidden state tensor.

  • setCellStateTensor: The function to set the cell state tensor.

PeepholeLongShortTermMemoryCell() 


RecurrentModels.PeepholeLongShortTermMemoryCell{inputDimensionSize: number, hiddenDimensionSize: number, learningRate: number}: function, WeightContainer, function

Parameters:

  • inputDimensionSize: The number of features it takes as inputs.

  • hiddenDimensionSize: The number of features it will produce.

  • learningRate: The speed at which the model learns. Recommended that the value is set between 0 to 1.

Returns:

  • Model: The model that is constructed using the set of parameters.

  • WeightContainer: The generated WeightContainer object.

  • reset: The function to reset the hidden state.

  • setCellStateTensor: The function to set the cell state tensor.

UncellModel() 


RecurrentModels.UncellModel{Model: function, reverse: boolean}: function

Parameters:

  • Model: The model to be given time sequence batching capabilities.

  • reverse: Set whether or not to train and predict from last element of the sequence to the first one.

Returns:

  • Model: The model that is constructed using the set of parameters.