If the workflow involves preprocessing the features, followed by model training, and finally model usage, then the preprocessing features step can be linked to the assumptions that we make when asking a question. For instance, the question can be, "Are these images of cats, knowing that cats have two ears, two eyes, a nose, a mouth, and whiskers?"

Our assumptions here are linked to how the images will be preprocessed to get the number of ears, eyes, noses, mouths, and whiskers. This data will be fed into the model during training so that we get answers.