Summary

This chapter covered CNNs. We reviewed core concepts such as neurons, layers, model architecture, and tensors to understand how to create effective CNNs.

You learned about the convolution operation and explored kernels and feature maps. We analyzed how to assemble a CNN, and then explored the different types of pooling layers and when to apply them.

You then learned about the stride operation and how padding is used to create extra space around images if needed. Then, we delved into the flattening layer and how it is able to convert data into a 1D array for the next layer. You put everything that you learned to the test in the final activity, as you were presented with several classification problems, including CIFAR-10 and even CIFAR-100.

In completing this chapter, you are now well on your way to being able to implement CNNs to confront image classification problems head-on and with confidence.

In the next chapter, you'll learn about pre-trained models and...