A Modified Drake Equation for Assessing Adversarial Risk to Machine Learning Models

TL;DR

This paper introduces a modified Drake Equation to estimate the number of successful adversarial attacks on machine learning models, providing a semi-quantitative benchmark for assessing risks in deployment.

Contribution

It adapts the Drake Equation formalism to quantify adversarial attack risks, offering a novel semi-quantitative framework for evaluating vulnerabilities in machine learning models.

Findings

Proposes a modified Drake Equation for adversarial risk estimation

Provides a semi-quantitative benchmark for model vulnerability assessment

Aims to assist industry in risk evaluation for deployed models

Abstract

Machine learning models present a risk of adversarial attack when deployed in production. Quantifying the contributing factors and uncertainties using empirical measures could assist the industry with assessing the risk of downloading and deploying common model types. This work proposes modifying the traditional Drake Equation's formalism to estimate the number of potentially successful adversarial attacks on a deployed model. The Drake Equation is famously used for parameterizing uncertainties and it has been used in many research fields outside of its original intentions to estimate the number of radio-capable extra-terrestrial civilizations. While previous work has outlined methods for discovering vulnerabilities in public model architectures, the proposed equation seeks to provide a semi-quantitative benchmark for evaluating and estimating the potential risk factors for adversarial attacks.