In the previous section, I mentioned the usefulness of data augmentation for small datasets. So, how do you use this during your training? The options are exposed to you using the  MLImageClassifier.ModelParameters structure, which you can pass an instance of when instantiating the classifier. One of the parameters is the OptionSet  CreateML.MLImageClassifier.ImageAugmentationOptions, which allows you to toggle various augmentation techniques on and off.

MLImageClassifier.ModelParameters also allows you to specify the maximum number of iterations, version of the feature extraction, and validation data. You can learn more about these on the official web page at https://developer.apple.com/documentation/create_ml/mlimageclassifier/modelparameters.