Controlling architecture generation with AutoKeras' AutoModel

Letting AutoKeras automagically figure out what architecture works best is great, but it can be time-consuming – unacceptably so at times.

Can we exert more control? Can we hint at which options work best for our particular problem? Can we meet AutoML halfway by providing a set of guidelines it must follow according to our prior knowledge or preference, but still give it enough leeway to experiment?

Yes, we can, and in this recipe, you'll learn how by utilizing a special feature in AutoKeras known as AutoModel!

How to do it…

Follow these steps to learn how to customize the search space of the NAS algorithm with AutoModel:

1. The first thing we need to do is import all the necessary dependencies:

   from autokeras import *
   from tensorflow.keras.datasets import fashion_mnist as fm
   from tensorflow.keras.models import load_model
   from tensorflow.keras.utils import *

2. Because we'll...