Hiding Behind Backdoors: Self-Obfuscation Against Generative Models

TL;DR

This paper introduces a self-obfuscation attack where attackers poison training data to make generative models obscure specific classes during inference, highlighting vulnerabilities in ML pipeline robustness.

Contribution

It describes, implements, and evaluates a generalized attack method targeting pre-processing models to obfuscate classes, raising awareness of architectural vulnerabilities.

Findings

Effective obfuscation of targeted classes during inference

Successful poisoning of training data to induce obfuscation

Demonstrated robustness challenges in ML architectures

Abstract

Attack vectors that compromise machine learning pipelines in the physical world have been demonstrated in recent research, from perturbations to architectural components. Building on this work, we illustrate the self-obfuscation attack: attackers target a pre-processing model in the system, and poison the training set of generative models to obfuscate a specific class during inference. Our contribution is to describe, implement and evaluate a generalized attack, in the hope of raising awareness regarding the challenge of architectural robustness within the machine learning community.