Keras Convolutional LSTM 2D Layer

2D Convolutional Long-Short Term Memory (LSTM) layer. Similar to a normal LSTM, but the input and recurrent transformations are both convolutional. Corresponds to the ConvLSTM2D Keras layer .

Options

Name prefix

The name prefix of the layer. The prefix is complemented by an index suffix to obtain a unique layer name. If this option is unchecked, the name prefix is derived from the layer type.

Input tensor

The tensor to use as input for the layer.

First hidden state tensor

The tensor to use as initial state for the first hidden state in case the corresponding port is connected.

Second hidden state tensor

The tensor to use as initial state for the second hidden state in case the corresponding port is connected.

Filters

Dimensionality of the output space.

Kernel size

A comma separated pair of integers that describes the size of the convolutional kernel.

Strides

A comma separated pair of positive integers that describes the stride of the network in each spatial dimension. A stride != 1 results in an output downsampled by a factor of 1/stride in the specific dimension. Note that any stride value != 1 is incompatible with specifying any dilation rate value != 1.

Padding

Same padding keeps the spatial size of the input intact (provided no stride or dilation is used) while valid padding reduces the size of the input.

Dilation rate

A comma separated pair of integers specifying the dilation rate to use for dilated convolution. Note that it is not supported to specify a stride value != 1 and a dilation rate value != 1.

Activation

The activation function to use on the input transformations.

Recurrent activation

The activation function to use for the recurrent step.

Use bias

If checked, a bias vector will be used.

Return sequences

Whether to return the last output in the output sequence or the full output sequence. If selected the output will have shape [time, height, width, channel] otherwise the output will have shape [height, width, channel] (for data format channel_last).

Return state

Whether to return the hidden states in addition to the layer output. If selected the layer returns three tensors, the normal output and the two hidden states of the layer. If sequences are returned, this also applies to the hidden states.

Go backwards

Whether to go backwards in time i.e. read the input sequence backwards.

Dropout

Fraction of the units to drop for the linear transformation of the input.

Recurrent dropout

Fraction of the units to drop for the linear transformation of the recurrent state.

Initializers

Kernel initializer

Initializer for the convolutional kernel used for the linear transformations of the input. See initializers for details on the available initializers.

Recurrent initializer

Initializer for the convolutional kernel used for the linear transformation of the recurrent connection. See initializers for details on the available initializers.

Bias initializer

Initializer for the bias vector (if a bias is used). See initializers for details on the available initializers.

Unit forget bias

If selected, add 1 to the bias of the forget gate at initialization. Use in combination with bias initializer zeros.

Regularizers

Kernel regularizer

Regularizer function applied to the convolutional kernel. See regularizers for details on the available regularizers.

Recurrent regularizer

Regularizer function applied to the convolutional kernel for the recurrent connection. See regularizers for details on the available regularizers.

Bias regularizer

Regularizer function applied to the bias vector. See regularizers for details on the available regularizers.

Activity regularizer

Regularizer function applied to the output of the layer i.e. its activation. See regularizers for details on the available regularizers.

Constraints

Kernel constraint

Constraint on the convolutional kernel for the input connection. See constraints for details on the available constraints.

Recurrent constraint

Constraint on the convolutional kernel for the recurrent connection. See constraints for details on the available constraints.

Bias constraint

Constraint on the bias vector. See constraints for details on the available constraints.

Views

This node has no views

Workflows

• No workflows found

Links

• KNIME Deep Learning Keras Integration

  www.knime.com/deeplearning/keras

Developers