An introduction to the colourful cast of characters¶

Where to begin with this motley crew of algorithms? Machine learning is rather like a mismatched football squad - you’ve got your by-the-book defenders (supervised learning), your chaotic strikers who just make it up as they go along (unsupervised learning), and that one player who somehow keeps scoring own goals (reinforcement learning on a bad day).

These digital brainboxes power everything from your dodgy sat-nav’s questionable route choices to the NHS’s attempts at predicting which of us will next clog up A&E after overdoing it at the local curry house. Some are discreet as MI5 (federated learning), while others gossip like a Wetherspoons regular after one too many pints (certain transductive models).

You’ve got the know-it-all swots (deductive learning), the lazy students who scrape by on minimal effort (semi-supervised learning), and the overefficient types who nick other people’s homework (transfer learning). Between them, they can identify your cat photos, recommend terrible telly, and occasionally - when they can be bothered - do something actually useful like detect cancer cells.