Prototype Based Classification from Hierarchy to Fairness

Mycal Tucker, Julie Shah

Keywords: prototypes, fairness, hierarchy, neural network, encoding

Abstract: Artificial neural nets can represent and classify many types of high-dimensional data but are often tailored to particular applications -- e.g., for ``fair'' or ``hierarchical'' classification. Once an architecture has been selected, it is often difficult for humans to adjust models for a new task; for example, a hierarchical classifier cannot be easily transformed into a fair classifier that shields a protected field. Our contribution in this work is a new neural network architecture, the concept subspace network (CSN), which generalizes existing specialized classifiers to produce a unified model capable of learning a spectrum of multi-concept relationships. We demonstrate that CSNs reproduce state-of-the-art results in fair classification when enforcing concept independence, may be transformed into hierarchical classifiers, or may even reconcile fairness and hierarchy within a single classifier. The CSN is inspired by and matches the performance of existing prototype-based classifiers that promote interpretability.

One-sentence Summary: A new neural network architecture with sets of prototypes allows us to represent and train concept relationships like fairness and hierarchy, all within one model.

Community Implementations: [ 2 code implementations](https://www.catalyzex.com/paper/arxiv:2205.13997/code)