Instance-Based Learning

noun

say it: “IN-stunss BAYST LUR-ning” /ˈɪnstəns ˈbeɪst ˈlɝnɪŋ/

A kind of machine learning that stores the training examples and makes each prediction by comparing a new example with them directly.

What is instance-based learning?

Instance-based learning keeps the training examples themselves instead of learning a formula, and compares each new example to them when it's time to guess, as k-nearest neighbors does. Training is instant, but every guess searches the stored examples, so guessing slows down as the data grows.

In a sentence

  • “Instance-based learning got slower as the shop's history grew to a million rows.”
  • “k-nearest neighbors is instance-based learning: its fit() only stores the examples.”

Level: Freelancer · More AI Basics terms at this level

Updated