Feature Attribution from First Principles

TL;DR

This paper introduces a new framework for feature attribution in machine learning that builds from simple models to complex ones, offering a more flexible and principled approach compared to axiomatic methods.

Contribution

It proposes a ground-up feature attribution framework based on simple indicator functions, recovering existing methods and deriving new closed-form solutions for deep networks.

Findings

Recovering existing attribution methods through atomic attributions

Deriving closed-form expressions for deep ReLU networks

Advancing evaluation metric optimization for feature attributions

Abstract

Feature attribution methods are a popular approach to explain the behavior of machine learning models. They assign importance scores to each input feature, quantifying their influence on the model's prediction. However, evaluating these methods empirically remains a significant challenge. To bypass this shortcoming, several prior works have proposed axiomatic frameworks that any feature attribution method should satisfy. In this work, we argue that such axioms are often too restrictive, and propose in response a new feature attribution framework, built from the ground up. Rather than imposing axioms, we start by defining attributions for the simplest possible models, i.e., indicator functions, and use these as building blocks for more complex models. We then show that one recovers several existing attribution methods, depending on the choice of atomic attribution. Subsequently, we derive closed-form expressions for attribution of deep ReLU networks, and take a step toward the optimization of evaluation metrics with respect to feature attributions.