Many Ways to be Right: Rashomon Sets for Concept-Based Neural Networks

TL;DR

This paper introduces Rashomon Concept Bottleneck Models, a method to efficiently learn multiple accurate neural networks that reason through different human-understandable concepts, revealing diverse decision processes for the same task.

Contribution

It presents a novel framework combining lightweight adapters and diversity regularization to uncover multiple concept-based solutions without retraining from scratch.

Findings

Networks exhibit diverse reasoning patterns for identical predictions.

The method efficiently constructs multiple models with different concept reliance.

Framework facilitates auditing and comparison of model reasoning.

Abstract

Modern neural networks rarely have a single way to be right. For many tasks, multiple models can achieve identical performance while relying on different features or reasoning patterns, a property known as the Rashomon Effect. However, uncovering this diversity in deep architectures is challenging as their continuous parameter spaces contain countless near-optimal solutions that are numerically distinct but often behaviorally similar. We introduce Rashomon Concept Bottleneck Models, a framework that learns multiple neural networks which are all accurate yet reason through distinct human-understandable concepts. By combining lightweight adapter modules with a diversity-regularized training objective, our method constructs a diverse set of deep concept-based models efficiently without retraining from scratch. The resulting networks provide fundamentally different reasoning processes for the same predictions, revealing how concept reliance and decision making vary across equally performing solutions. Our framework enables systematic exploration of data-driven reasoning diversity in deep models, offering a new mechanism for auditing, comparison, and alignment across equally accurate solutions.