Machine Learning · · 7 min read
Understanding Machine Learning
An accessible overview of how machines learn from data, and what that means for the people deploying them.
Machine learning is a way of building software where behaviour is learned from examples instead of written as explicit rules.
Three common forms
- Supervised learning: learning from labelled examples
- Unsupervised learning: finding structure in unlabelled data
- Reinforcement learning: learning through feedback on actions
Generalisation is the point
A model is only useful if it performs well on data it has not seen. Most of the discipline of machine learning — validation sets, regularisation, careful evaluation — exists to protect against models that simply memorise.
Machine.sa editorial