Nearest-Neighbor Sample Compression: Efficiency, Consistency, Infinite Dimensions

TL;DR

This paper proves the Bayes-consistency of a 1-nearest-neighbor sample compression algorithm in finite and certain infinite-dimensional metric spaces, highlighting its statistical and computational advantages.

Contribution

It establishes the first consistency result for an efficient nearest-neighbor sample compression scheme in finite doubling dimension spaces and extends this to some infinite-dimensional settings.

Findings

Proves strong Bayes-consistency in finite doubling dimension spaces.

Shows the algorithm remains consistent in some infinite-dimensional spaces.

Highlights the algorithm's efficiency and empirical generalization bounds.

Abstract

We examine the Bayes-consistency of a recently proposed 1-nearest-neighbor-based multiclass learning algorithm. This algorithm is derived from sample compression bounds and enjoys the statistical advantages of tight, fully empirical generalization bounds, as well as the algorithmic advantages of a faster runtime and memory savings. We prove that this algorithm is strongly Bayes-consistent in metric spaces with finite doubling dimension --- the first consistency result for an efficient nearest-neighbor sample compression scheme. Rather surprisingly, we discover that this algorithm continues to be Bayes-consistent even in a certain infinite-dimensional setting, in which the basic measure-theoretic conditions on which classic consistency proofs hinge are violated. This is all the more surprising, since it is known that $k$-NN is not Bayes-consistent in this setting. We pose several challenging open problems for future research.