Chapter 5. Getting started with neural networks

 

This chapter covers

  • Introducing the fundamentals of artificial neural networks
  • Teaching a network to recognize handwritten digits
  • Creating neural networks by composing layers
  • Understanding how neural networks learn from data
  • Implementing a simple neural network from scratch

This chapter introduces the core notions of artificial neural networks (ANNs), a class of algorithms central to modern-day deep learning. The history of artificial neural networks is a surprisingly old one, dating back to the early 1940s. It took many decades for its applications to become vast successes in many areas, but the basic ideas remain in effect.

5.1. A simple use case: classifying handwritten digits

5.2. The basics of neural networks

5.3. Feed-forward networks

5.4. How good are our predictions? Loss functions and optimization

5.5. Training a neural network step-by-step in Python

5.6. Summary