CNN architecture summary (IMAGE)
Caption
CNN architecture summary: The first dimension in all the layers “?” refers to the batch size. It is left as an unknown or unspecified variable within the network architecture so that it can be chosen during training. All layers before flattening had four dimensions. The second and third dimensions corresponded to the matrix dimensions. Before the convolution, the input matrix had the dimensions 8 × 11, as previously explained. The last dimension corresponded to the channel dimension. In the Input Layer, this adopted a value of 1 since we merely inputted a two-dimensional matrix. Conv2D used 32 filters and generated 32 activations from the original one-dimensional input. Hence, the last dimension expanded to 32 after Conv2D. After flattening the data, we were left with two dimensions: the first was the batch size as before, and the second was the length of the array. Finally, we coded a common dense layer and five different layers to predict each of the model parameters. The last layer simply concatenated the prediction of each of the model parameters into one array.
Credit
Brigo, D., et al
Usage Restrictions
Credit must be given to the creator. Only noncommercial uses of the work are permitted. No derivatives or adaptations of the work are permitted.
License
CC BY-NC-ND