TimeCNNNetwork¶
- class TimeCNNNetwork(n_layers=2, kernel_size=7, n_filters=None, avg_pool_size=3, activation='sigmoid', padding='valid', strides=1, strides_pooling=None, dilation_rate=1, use_bias=True)[source]¶
Establish the network structure for a CNN.
Adapted from the implementation used in [1].
- Parameters:
- n_layersint, default = 2
The number of convolution layers in the network.
- kernel_sizeint or list of int, default = 7
Kernel size of convolution layers, if not a list, the same kernel size is used for all layer, len(list) should be n_layers.
- n_filtersint or list of int, default = [6, 12]
Number of filters for each convolution layer, if not a list, the same n_filters is used in all layers.
- avg_pool_sizeint or list of int, default = 3
The size of the average pooling layer, if not a list, the same max pooling size is used for all convolution layer.
- activationstr or list of str, default = “sigmoid”
Keras activation function used in the model for each layer, if not a list, the same activation is used for all layers.
- paddingstr or list of str, default = “valid”
The method of padding in convolution layers, if not a list, the same padding used for all convolution layers.
- stridesint or list of int, default = 1
The strides of kernels in the convolution and max pooling layers, if not a list, the same strides are used for all layers.
- strides_poolingint or list of int, default = None
Strides for the pooling layers. If None, defaults to pool_size. If not a list, the same strides are used for all pooling layers.
- dilation_rateint or list of int, default = 1
The dilation rate of the convolution layers, if not a list, the same dilation rate is used all over the network.
- use_biasbool or list of bool, default = True
Condition on whether or not to use bias values for convolution layers, if not a list, the same condition is used for all layers.
Notes
Adapted from source code https://github.com/hfawaz/dl-4-tsc/blob/master/classifiers/cnn.py
References
[1]Zhao et al. Convolutional neural networks for time series classification,
Journal of Systems Engineering and Electronics 28(1), 162–169, 2017
Methods
build_network(input_shape, **kwargs)Construct a network and return its input and output layers.