crested.tl.zoo.simple_convnet

Contents

crested.tl.zoo.simple_convnet#

crested.tl.zoo.simple_convnet(seq_len, num_classes, num_conv_blocks=3, num_dense_blocks=2, residual=False, first_activation='exponential', activation='swish', output_activation='softplus', normalization='batch', first_filters=192, filters=256, first_kernel_size=13, kernel_size=7, first_pool_size=8, pool_size=2, conv_dropout=0.1, dense_dropout=0.3, flatten=True, dense_size=256, bottleneck=8)#

Construct a Simple ConvNet with standard convolutional and dense blocks.

Used as a baseline model for regression or classification tasks.

Parameters:

  • seq_len (int) – Width of the input region.

  • num_classes (int) – Number of classes to predict.

  • num_conv_blocks (int (default: 3)) – Number of convolutional blocks.

  • num_dense_blocks (int (default: 2)) – Number of dense blocks.

  • residual (bool (default: False)) – Whether to use residual connections.

  • first_activation (str (default: 'exponential')) – Activation function for the first convolutional block.

  • activation (str (default: 'swish')) – Activation function for subsequent convolutional and dense blocks.

  • output_activation (str (default: 'softplus')) – Activation function for the output layer.

  • normalization (str (default: 'batch')) – Type of normalization (‘batch’ or ‘layer’).

  • first_filters (int (default: 192)) – Number of filters in the first convolutional block.

  • filters (int (default: 256)) – Number of filters in subsequent convolutional blocks.

  • first_kernel_size (int (default: 13)) – Size of the kernel in the first convolutional block.

  • kernel_size (int (default: 7)) – Size of the kernel in subsequent convolutional blocks.

  • first_pool_size (int (default: 8)) – Size of the pooling kernel in the first convolutional block.

  • pool_size (int (default: 2)) – Size of the pooling kernel in subsequent convolutional blocks.

  • conv_dropout (float (default: 0.1)) – Dropout rate for the convolutional layers.

  • dense_dropout (float (default: 0.3)) – Dropout rate for the dense layers.

  • flatten (bool (default: True)) – Whether to flatten the output before dense layers.

  • dense_size (int (default: 256)) – Number of neurons in the dense layers.

  • bottleneck (int (default: 8)) – Size of the bottleneck layer.

Return type:

Model

Returns:

A Keras model.

Contents