Number of transposed convolution filters to use.
Number of rows in the transposed convolution kernel.
Number of columns in the transposed convolution kernel.
Initialization method for the weights of the layer. Default is Xavier. You can also pass in corresponding string representations such as 'glorot_uniform' or 'normal', etc. for simple init methods in the factory method.
Activation function to use. Default is null. You can also pass in corresponding string representations such as 'relu' or 'sigmoid', etc. for simple activations in the factory method.
Int array of length 2. The step of the convolution in the height and width dimension. Also called strides elsewhere. Default is (1, 1).
Format of input data. Please use DataFormat.NCHW (dimOrdering='th').
An instance of Regularizer, (eg. L1 or L2 regularization), applied to the input weights matrices. Default is null.
An instance of Regularizer, applied to the bias. Default is null.
Whether to include a bias (i.e. make the layer affine rather than linear). Default is true.
A Single Shape, does not include the batch dimension.
Activation function to use.
Activation function to use. Default is null. You can also pass in corresponding string representations such as 'relu' or 'sigmoid', etc. for simple activations in the factory method.
Whether to include a bias (i.
Whether to include a bias (i.e. make the layer affine rather than linear). Default is true.
Format of input data.
Format of input data. Please use DataFormat.NCHW (dimOrdering='th').
Build graph: some other modules point to current module
Build graph: some other modules point to current module
upstream variables
Variable containing current module
Initialization method for the weights of the layer.
Initialization method for the weights of the layer. Default is Xavier. You can also pass in corresponding string representations such as 'glorot_uniform' or 'normal', etc. for simple init methods in the factory method.
A Single Shape, does not include the batch dimension.
A Single Shape, does not include the batch dimension.
Number of columns in the transposed convolution kernel.
Number of columns in the transposed convolution kernel.
Number of transposed convolution filters to use.
Number of transposed convolution filters to use.
Number of rows in the transposed convolution kernel.
Number of rows in the transposed convolution kernel.
Int array of length 2.
Int array of length 2. The step of the convolution in the height and width dimension. Also called strides elsewhere. Default is (1, 1).
(Since version 0.3.0) please use recommended saveModule(path, overWrite)
Transposed convolution operator for filtering windows of 2-D inputs. The need for transposed convolutions generally arises from the desire to use a transformation going in the opposite direction of a normal convolution, i.e., from something that has the shape of the output of some convolution to something that has the shape of its input while maintaining a connectivity pattern that is compatible with said convolution. Data format currently supported for this layer is DataFormat.NCHW (dimOrdering='th'). Border mode currently supported for this layer is 'valid'. You can also use Deconv2D as an alias of this layer. The input of this layer should be 4D.
When using this layer as the first layer in a model, you need to provide the argument inputShape (a Single Shape, does not include the batch dimension). e.g. inputShape=Shape(3, 128, 128) for 128x128 RGB pictures.
The numeric type of parameter(e.g. weight, bias). Only support float/double now.