Optimal Explanations of Linear Models

TL;DR

This paper introduces a unified optimization framework for generating explanations of linear models, enabling a systematic tradeoff analysis between interpretability and accuracy.

Contribution

It proposes a general method to decompose linear models into interpretable components using exact algorithms and heuristics, advancing model interpretability research.

Findings

Develops an optimization-based explanation method for linear models.

Provides algorithms for efficient computation of explanations.

Analyzes the tradeoff between interpretability and predictive accuracy.

Abstract

When predictive models are used to support complex and important decisions, the ability to explain a model's reasoning can increase trust, expose hidden biases, and reduce vulnerability to adversarial attacks. However, attempts at interpreting models are often ad hoc and application-specific, and the concept of interpretability itself is not well-defined. We propose a general optimization framework to create explanations for linear models. Our methodology decomposes a linear model into a sequence of models of increasing complexity using coordinate updates on the coefficients. Computing this decomposition optimally is a difficult optimization problem for which we propose exact algorithms and scalable heuristics. By solving this problem, we can derive a parametrized family of interpretability metrics for linear models that generalizes typical proxies, and study the tradeoff between interpretability and predictive accuracy.