theanets.layers.convolution.Convolution

class theanets.layers.convolution.Convolution(filter_shape, stride=(1, 1), border_mode='valid', **kwargs)

Convolution layers convolve filters over the input arrays.

Parameters:

filter_shape : (int, int)

Shape of the convolution filters for this layer.

stride : (int, int), optional

Apply convolutions with this stride; i.e., skip this many samples between convolutions. Defaults to (1, 1)—that is, no skipping.

border_mode : str, optional

Compute convolutions with this border mode. Defaults to ‘valid’.

__init__(filter_shape, stride=(1, 1), border_mode='valid', **kwargs)

Methods

__init__(filter_shape[, stride, border_mode])
add_conv_weights(name[, mean, std, sparsity])	Add a convolutional weight array to this layer’s parameters.

Attributes

input_size	For networks with one input, get the input size.
num_params	Total number of learnable parameters in this layer.
params	A list of all parameters in this layer.

add_conv_weights(name, mean=0, std=None, sparsity=0)

Add a convolutional weight array to this layer’s parameters.

Parameters:

name : str

Name of the parameter to add.

mean : float, optional

Mean value for randomly-initialized weights. Defaults to 0.

std : float, optional

Standard deviation of initial matrix values. Defaults to \(1 / sqrt(n_i + n_o)\).

sparsity : float, optional

Fraction of weights to set to zero. Defaults to 0.