SurFree: a fast surrogate-free black-box attack

11/25/2020
by   Thibault Maho, et al.
0

Machine learning classifiers are critically prone to evasion attacks. Adversarial examples are slightly modified inputs that are then misclassified, while remaining perceptively close to their originals. Last couple of years have witnessed a striking decrease in the amount of queries a black box attack submits to the target classifier, in order to forge adversarials. This particularly concerns the black-box score-based setup, where the attacker has access to top predicted probabilites: the amount of queries went from to millions of to less than a thousand. This paper presents SurFree, a geometrical approach that achieves a similar drastic reduction in the amount of queries in the hardest setup: black box decision-based attacks (only the top-1 label is available). We first highlight that the most recent attacks in that setup, HSJA, QEBA and GeoDA all perform costly gradient surrogate estimations. SurFree proposes to bypass these, by instead focusing on careful trials along diverse directions, guided by precise indications of geometrical properties of the classifier decision boundaries. We motivate this geometric approach before performing a head-to-head comparison with previous attacks with the amount of queries as a first class citizen. We exhibit a faster distortion decay under low query amounts (few hundreds to a thousand), while remaining competitive at higher query budgets.

READ FULL TEXT
research
03/13/2022

Query-Efficient Black-box Adversarial Attacks Guided by a Transfer-based Prior

Adversarial attacks have been extensively studied in recent years since ...
research
05/30/2018

AutoZOOM: Autoencoder-based Zeroth Order Optimization Method for Attacking Black-box Neural Networks

Recent studies have shown that adversarial examples in state-of-the-art ...
research
05/16/2019

Parsimonious Black-Box Adversarial Attacks via Efficient Combinatorial Optimization

Solving for adversarial examples with projected gradient descent has bee...
research
10/14/2020

Explain2Attack: Text Adversarial Attacks via Cross-Domain Interpretability

Training robust deep learning models for down-stream tasks is a critical...
research
06/05/2023

Evading Black-box Classifiers Without Breaking Eggs

Decision-based evasion attacks repeatedly query a black-box classifier t...
research
12/23/2017

Query-limited Black-box Attacks to Classifiers

We study black-box attacks on machine learning classifiers where each qu...
research
08/24/2021

Transient Execution of Non-Canonical Accesses

Recent years have brought microarchitectural security intothe spotlight,...

Please sign up or login with your details

Forgot password? Click here to reset