Chapter 2 · Machine learning basics
Labels and supervised learning
In supervised learning every example comes with the right answer, and the model learns to reproduce it.
Preparing labels and supervised learning
Chapter 2 · Machine learning basics
In supervised learning every example comes with the right answer, and the model learns to reproduce it.
Each example arrives with its answer on the back. The learner compares its own guess with the answer and improves.
Gather many examples, such as photos, emails or house details.
Tap a dot to label it A (violet), then B (coral), then clear. As soon as both classes have a label, the map shades to show what the model would say everywhere.
Label at least one dot A and one dot B to start. Teaching version: the model just compares each spot with the average of every class.
What makes learning 'supervised'?
If many labels in the training data are wrong, the model will…
Which is a label for a spam filter?