Computing Class Hierarchies from Classifiers
This provides a method for interpretability and other potential uses in neural networks, but is incremental as it builds on existing ideas of using confusion matrices.
The paper tackles the problem of automatically constructing class hierarchies from classifier confusion matrices, and demonstrates that their algorithm produces surprisingly good hierarchies for several neural network models across different domains.
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.