Detailed Documentation

Data Preprocessing Scripts

Data Normalization

Training Neural Networks

Brunton Test

Timing Predictors and other tests

Using a custom model

To train more complex models, you can write your own:

1. Write your module SI_Toolkit_ASF/ToolkitCustomization/Modules/[module_name].py

• Write a class that inherits from tf.keras.Model
• The class must have a __init__(self) function and a call(self, x) function. x is a tensor containing the input to the network. Refer to https://keras.io/api/models/model/ for more information on how to write your own model.

2. For training, follow the steps outlined above.

• Use your custom model by specifying --net-name Custom-[module_name]-[class_name]