Transferability in Machine Learning: from Phenomena to Black-Box Attacks using Adversarial Samples

TLDR

Machine‑learning models are vulnerable to adversarial examples that can transfer across different architectures, allowing an attacker to train a substitute model and use the victim as an oracle to generate attacks with minimal information. We aim to improve transfer‑attack efficiency by extending recent techniques with reservoir sampling and to investigate new substitute‑victim model pairs, notably SVMs and decision trees. Our method applies reservoir sampling during substitute‑model training and uses the victim model as an oracle to label synthetic data, allowing attacks with only 800 queries. The attacks achieve 96.19 % and 88.94 % misclassification rates on Amazon and Google commercial classifiers, demonstrating that black‑box attacks can compromise diverse machine‑learning systems.

Abstract

Many machine learning models are vulnerable to adversarial examples: inputs that are specially crafted to cause a machine learning model to produce an incorrect output. Adversarial examples that affect one model often affect another model, even if the two models have different architectures or were trained on different training sets, so long as both models were trained to perform the same task. An attacker may therefore train their own substitute model, craft adversarial examples against the substitute, and transfer them to a victim model, with very little information about the victim. Recent work has further developed a technique that uses the victim model as an oracle to label a synthetic training set for the substitute, so the attacker need not even collect a training set to mount the attack. We extend these recent techniques using reservoir sampling to greatly enhance the efficiency of the training procedure for the substitute model. We introduce new transferability attacks between previously unexplored (substitute, victim) pairs of machine learning model classes, most notably SVMs and decision trees. We demonstrate our attacks on two commercial machine learning classification systems from Amazon (96.19% misclassification rate) and Google (88.94%) using only 800 queries of the victim model, thereby showing that existing machine learning approaches are in general vulnerable to systematic black-box attacks regardless of their structure.