Running image classification and analyzing the results

This example demonstrates the use of this FL framework for image classification tasks. We will use a famous image dataset, CIFAR-10 (URL: https://www.cs.toronto.edu/~kriz/cifar.html), to show how an ML model grows through the FL process over time. However, this example is only given for the purposes of using the FL system we have discussed so far and is not focused on maximizing the performance of the image classification task.

Preparing the CIFAR-10 dataset

The following is the information required related to the dataset size, the training and test data, the number of classes, and the image size:

• Dataset size: 60,000 images
• Training data: 50,000 images
• Test data: 10,000 images
• Number of classes: 10 (airplane, automobile, bird, cat, deer, dog, frog, horse, ship, and truck)
• Each class has 6,000 images
• Image size: 32x32 pixels, in color

Figure 6.11 shows a collection of sample pictures of...