Table of Contents

🎓 Intended learning outcomes

At the end of this lesson, students are expected to:

• describe a neural network and express it mathematically
• know the terms layer, activation function, activation, preactivation, weight, bias, width, and depth
• know the purpose of the common activation functions
• relate neural networks to their biological connections
• describe end-to-end learning
• understand and describe how each layer of a neural network transforms the data
• explain compositional learning
• explain why neural networks are universal function approximators and what that means
• understand how network size and regularization determine a neural networks expressiveness

In this lesson, we will discuss the basics of neural networks.

Imagine you’ve just started a new job at PostNord working as an ML engineer. You’ve been tasked with developing an automatic method to read postcodes from letters and packages. The rest of the team is working on a system to extract images of the postcode, and your job is to classify the individual digits so it can be automatically entered into the computer system and routed to the correct regional post center. Let’s see how you can use neural networks to help with this task.