AEFCNNetwork¶
- class AEFCNNetwork(latent_space_dim=128, temporal_latent_space=False, n_layers=3, n_filters=None, kernel_size=None, dilation_rate=1, strides=1, padding='same', activation='relu', use_bias=True)[source]¶
Establish the network structure for a AE-FCN.
Auto-Encoder based Fully Convolutional Netwwork (AE-FCN), adapted from the implementation used in [1].
- Parameters:
- latent_space_dimint, default = 128
Dimension of the auto-encoder’s latent space.
- temporal_latent_spacebool, default = False
Flag to choose whether the latent space is an MTS or Euclidean space.
- n_layersint, default = 3
Number of convolution layers.
- n_filtersint or list of int, default = [128,256,128]
Number of filters used in convolution layers.
- kernel_sizeint or list of int, default = [8,5,3]
Size of convolution kernel.
- dilation_rateint or list of int, default = 1
The dilation rate for convolution.
- stridesint or list of int, default = 1
The strides of the convolution filter.
- paddingstr or list of str, default = “same”
The type of padding used for convolution.
- activationstr or list of str, default = “relu”
Activation used after the convolution.
- use_biasbool or list of bool, default = True
Whether or not ot use bias in convolution.
Notes
Adapted from the implementation from Fawaz et. al https://github.com/hfawaz/dl-4-tsc/blob/master/classifiers/fcn.py
References
[1]Network originally defined in:
- @inproceedings{wang2017time,
- title={Time series classification from scratch with deep neural networks:
A strong baseline},
author={Wang, Zhiguang and Yan, Weizhong and Oates, Tim}, booktitle={2017 International joint conference on neural networks (IJCNN)}, pages={1578–1585}, year={2017}, organization={IEEE}
}
Methods
build_network(input_shape, **kwargs)Construct a network and return its input and output layers.