Occam's razor

Occam's razor (variously spelt Ockam, Ocham) is the principle that the simplest explanation is often the best or correct one. It traces back to a 13th century monk William of Ockham, and is often stated as in Latin as "pluralitas non est ponenda sine necessitate", which trabslates to “plurality (or entities) should not be posited without necessity.” In AI this principle is often used to improve the generalisation after machine learning , for example pruning decision trees or the pinch points in deep neural networks. It is closely related to issues of overfitting as models with too many free parameters often pick up irrelevant details of particular training examples.

Used on page 91