In a single sentence, nearest neighbor classifiers are defined by their characteristic of classifying unlabeled examples by assigning them the class of similar labeled examples. This is analogous to the dining experience described in the chapter introduction, in which a person identifies new foods through comparison to those previously encountered. With nearest neighbor classification, computers apply a human-like ability to recall past experiences to make conclusions about current circumstances. Despite the simplicity of this idea, nearest neighbor methods are extremely powerful. They have been used successfully for:
Computer vision applications, including optical character recognition and facial recognition in both still images and video
Recommendation systems that predict whether a person will enjoy a movie or song
Identifying patterns in genetic data to detect specific proteins or diseases
In general, nearest neighbor classifiers are well suited...
