PE1-6.3 Neural Networks
You can describe models of a neuron and neural networks as directed graphs, distinguish feedforward from feedback architectures, and classify learning processes and learning tasks.
Viewing a network as a directed graph is the framing that makes architecture a design choice rather than a diagram to memorise — nodes compute, edges carry weights, and feedback edges give the network memory. The learning-process taxonomy matters more than the biological analogy: what changes the weights, and what signal tells it to change. This unit is compact because Unit 4 does the algorithms; treat it as the vocabulary layer.
Work through these
What a neural network is; the human brain as motivation
What these systems are and the biological analogy behind them. The analogy is loose, and treating it as loose keeps expectations honest.
Models of a neuron
The single unit and the ways of modelling it. Everything else is many of these arranged and trained.
Neural networks as a directed graph
Drawing a network as a graph, which makes the flow of signals explicit. It is the notation the architecture item below uses.
Feedback and network architectures
How units are connected, including loops. Whether a network has feedback changes what it can represent and how it must be trained.
Learning processes
How the weights change, which is where learning happens. Several rules are variations on moving towards a better answer.
Learning tasks
What these systems are asked to do, from classification to control. It connects the machinery above to the applications in the last topic.
Sign in to keep your progress.
Free resources
We haven't checked most of these for screen reader use yet.
Links last checked 30 Aug 2026.
Stuck here?
Ask a mentor. A real person answers, and they can see exactly which topic you're on. Usually within a couple of working days.
Checking your session…
Topics shown in module order.