To complete this project, we will use the MNIST digit dataset, which is available in the TensorFlow datasets library (https://www.tensorflow.org/guide/datasets). It consists of images of handwritten digits from 0 to 9. The training dataset has 60,000 images and the testing set has 10,000 images. Some of the images in the dataset are as follows:
If we take a look at TensorFlow Lite tutorials, we will see that the focus is on using pre-trained models such as Mobilenet or retraining the existing ones. However, none of these tutorials talk about building new models, which is something we will be doing here.
Note that we specifically choose a simple model because at the time of writing this book, TensorFlow Lite doesn't have adequate support for all types of complex models
We will use categorical cross entropy as the loss...
