DCNNNetwork

class DCNNNetwork(latent_space_dim=128, n_layers=4, kernel_size=3, activation='relu', n_filters=None, dilation_rate=None, padding='causal')[source]

Establish the network structure for a DCNN-Model.

Dilated Convolutional Neural Network based Model for low-rank embeddings.

Parameters:
latent_space_dim: int, default=128

Dimension of the models’s latent space.

n_layers: int, default=4

Number of convolution layers.

kernel_size: Union[int, List[int]], default=3

Size of the 1D Convolutional Kernel. Defaults to a list of three’s for n_layers elements.

activation: Union[str, List[str]], default=”relu”

The activation function used by convolution layers. Defaults to a list of “relu” for n_layers elements.

n_filters: Union[int, List[int]], default=None

Number of filters used in convolution layers. Defaults to a list of multiple’s of 32 for n_layers elements.

dilation_rate: Union[int, List[int]], default=None

The dilation rate for convolution. Defaults to a list of powers of 2 for n_layers elements.

padding: Union[str, List[str]], default=”causal”

Padding to be used in each DCNN Layer. Defaults to a list of causal paddings for n_layers elements.

References

[1]

Franceschi, J. Y., Dieuleveut, A., & Jaggi, M. (2019).

Unsupervised scalable representation learning for multivariate time series. Advances in neural information processing systems, 32.

Methods

build_network(input_shape)

Construct a network and return its input and output layers.

build_network(input_shape)[source]

Construct a network and return its input and output layers.

Parameters:
input_shapetuple of shape = (n_timepoints (m), n_channels (d))

The shape of the data fed into the input layer.

Returns:
modela keras Model.