Computing Class Hierarchies from Classifiers

TL;DR

This paper introduces a novel algorithm that automatically derives class hierarchies from neural network confusion matrices, enhancing interpretability and understanding of classifiers across various domains.

Contribution

The paper presents a new method for extracting class hierarchies directly from confusion matrices of neural networks, applicable to diverse models and datasets.

Findings

Produced effective hierarchies for well-known neural networks on CIFAR-10

Successfully applied to language and music genre classifiers

Hierarchies improve interpretability of neural network decisions

Abstract

A class or taxonomic hierarchy is often manually constructed, and part of our knowledge about the world. In this paper, we propose a novel algorithm for automatically acquiring a class hierarchy from a classifier which is often a large neural network these days. The information that we need from a classifier is its confusion matrix which contains, for each pair of base classes, the number of errors the classifier makes by mistaking one for another. Our algorithm produces surprisingly good hierarchies for some well-known deep neural network models trained on the CIFAR-10 dataset, a neural network model for predicting the native language of a non-native English speaker, a neural network model for detecting the language of a written text, and a classifier for identifying music genre. In the literature, such class hierarchies have been used to provide interpretability to the neural networks. We also discuss some other potential uses of the acquired hierarchies.