Learning a metric for class-conditional KNN

TL;DR

This paper introduces Class Conditional Metric Learning (CCML), a novel approach that optimizes class-aware similarity measures to improve KNN-based classification and retrieval across diverse datasets.

Contribution

It proposes a new metric learning method that captures perceptual similarity and can be tuned for different levels of locality, extending NBNN's applicability beyond image data.

Findings

CCML outperforms existing learned metrics on various datasets.

The method effectively captures class-based perceptual similarity.

It improves classification and retrieval accuracy in both image and non-image domains.

Abstract

Naive Bayes Nearest Neighbour (NBNN) is a simple and effective framework which addresses many of the pitfalls of K-Nearest Neighbour (KNN) classification. It has yielded competitive results on several computer vision benchmarks. Its central tenet is that during NN search, a query is not compared to every example in a database, ignoring class information. Instead, NN searches are performed within each class, generating a score per class. A key problem with NN techniques, including NBNN, is that they fail when the data representation does not capture perceptual (e.g.~class-based) similarity. NBNN circumvents this by using independent engineered descriptors (e.g.~SIFT). To extend its applicability outside of image-based domains, we propose to learn a metric which captures perceptual similarity. Similar to how Neighbourhood Components Analysis optimizes a differentiable form of KNN classification, we propose "Class Conditional" metric learning (CCML), which optimizes a soft form of the NBNN selection rule. Typical metric learning algorithms learn either a global or local metric. However, our proposed method can be adjusted to a particular level of locality by tuning a single parameter. An empirical evaluation on classification and retrieval tasks demonstrates that our proposed method clearly outperforms existing learned distance metrics across a variety of image and non-image datasets.