AI & Machine Learning · Intermediate

K-Nearest Neighbors (k-NN)

A non-parametric algorithm that predicts by majority vote or average of the k closest training examples.

Definition

k-NN stores the training data and, at prediction time, finds the k nearest points by a distance metric. It needs no training phase but inference is slow on large datasets and it is sensitive to feature scaling and high dimensionality. Approximate nearest neighbor search applies the same idea to vector databases.

Related terms