DeepFool: A Simple and Accurate Method to Fool Deep Neural Networks
By Seyed-Mohsen Moosavi-Dezfooli, Alhussein Fawzi, Pascal Frossard. 2015
5,469 citations recorded by the source corpus.
State-of-the-art deep neural networks have achieved impressive results on many image classification tasks. However, these same architectures have been shown to be unstable to small, well sought, perturbations of the images. Despite the importance of this phenomenon, no effective methods have been proposed to accurately compute the robustness of state-of-the-art deep classifiers to such perturbations on large-scale datasets. In this paper, we fill this gap and propose the DeepFool algorithm to efficiently compute perturbations that fool deep networks, and thus reliably quantify the robustness of these classifiers. Extensive experimental results show that our approach outperforms recent methods in the task of computing adversarial perturbations and making classifiers more robust.
Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Safety, Robustness & Interpretability
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Percy Liang
- Yoshua Bengio
- Mohit Bansal
- Quoc V. Le
- Bernt Schiele
- Jianfeng Gao
- Sergey Levine
- Nicholas Carlini
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Dacheng Tao
- Trevor Darrell
- Ming-Hsuan Yang
- Bernt Schiele
- Xiaogang Wang
- Alan L. Yuille
- Jitendra Malik
In Safety, Robustness & Interpretability
- Nicholas Carlini
- Matthias Hein
- Nicolas Papernot
- Aleksander Madry
- Bo Li
- Cho-Jui Hsieh
- Florian Tramèr
- Huan Zhang
Citations in the corpus
- Repurposing Pretrained Models for Robust Out-of-domain Few-Shot Learning (2021, background): “Deep NNs are sensitive to adversarial attacks (Goodfellow et al., 2015; Madry et al., 2018; Moosavi-Dezfooli et al., 2016).”
- Adversarial Example Games (2020, methodology): “For example, in the popular whitebox threat model, the attacker is assumed to have full access to the model f ’s parameters and outputs [62, 30, 51, 53, 11, 16, 17].”
- Towards the Science of Security and Privacy in Machine Learning (2016, background): “Follow-up work reduced the size of a perturbation r according to different metrics [53], [66].”
- Towards Deep Learning Models Resistant to Adversarial Attacks (2017, background): “Computer vision presents a particularly striking challenge: very small changes to the input image can fool state-of-the-art neural networks with high probability (Szegedy et al., 2013; Goodfellow et al., 2014; Nguyen et al., 2015; Sharif et al., 2016; Moosavi-Dezfooli et al., 2016).”
- Understanding Black-box Predictions via Influence Functions (2017, methodology): “Our work connects the literature on training-set attacks with work on “adversarial examples” (Goodfel-low et al., 2015; Moosavi-Dezfooli et al., 2016), visually-imperceptible perturbations on test inputs.”
- Wild Patterns: Ten Years After the Rise of Adversarial Machine Learning (2017, background): “The rationale in this case is to maximize the con-fidence assigned to the wrong target class f k , while minimizing the probability of correct classification [4, 6].”
- Deep Variational Information Bottleneck (2017, background): “6 There are also other adversaries that don’t fall as cleanly into those categories, such as “fooling images” from Nguyen et al. (2015), which remove the human perceptual constraint, generating regular geometric patterns or noise patterns that networks confidently classify as natural images; and the”
- Attack of the Tails: Yes, You Really Can Backdoor Federated Learning (2020, background): “We establish theoretically that if a model is vulnerable to adversarial examples, such as the ones presented in [28, 29, 30, 31, 32], then, under mild conditions, backdoor attacks are unavoidable.”
- SentiNet: Detecting Localized Universal Attacks Against Deep Learning Systems (2018, methodology): “by adding small gradient perturbations to an image that is imperceptible to humans. Numerous works have built on this approach; some notable works include the Fast Gradient Sign Method [18], DeepFool [35], and Universal Adversarial Perturbations [34]. This area of research can be categorized as a ca”
- Stronger data poisoning attacks break data sanitization defenses (2018, background): “perceptible way and yet fool state-of-the-art neural network image classiers (Szegedy et al., 2014; Goodfellow et al., 2015; Carlini et al., 2016; Kurakin et al., 2016; Papernot et al., 2016a, 2017; Moosavi-Dezfooli et al., 2016). Designing models that are robust to such attacks, as well as coming u”
- Decision-Based Adversarial Attacks: Reliable Attacks Against Black-Box Machine Learning Models (2017, methodology): “• DeepFool.”
- A Rotation and a Translation Suffice: Fooling CNNs with Simple Transformations (2017, methodology): “There is a long line of work (Szegedy et al., 2013; Goodfellow et al., 2014; Kurakin et al., 2016a;b; Sharif et al., 2016; Moosavi-Dezfooli et al., 2016; Carlini & Wagner, 2016; Papernot et al., 2017; Madry et al., 2017; Athalye et al., 2017) on methods for constructing adversarial perturbations Equ”
- Ensemble Adversarial Training: Attacks and Defenses (2017, background): “, perturbations obtained by taking a single step along the gradient of the model’s loss [5, 8], but remain vulnerable to stronger attacks that iteratively maximize the model’s loss [10, 8].”
- Towards Evaluating the Robustness of Neural Networks (2016, methodology): “The exact formulation used is rather sophisticated; interested readers should refer to the original work [34].”
- Contamination Attacks and Mitigation in Multi-Party Machine Learning (2018, background): “Our attacks exploit misaligned goals between parties in multi-party machine learning as opposed to exploiting vulnerabilities within the model itself, such as with adversarial examples [7, 13, 24, 34, 30].”
- Property Inference Attacks on Fully Connected Neural Networks using Permutation Invariant Representations (2018, background): “Beyond data inference, MLmodels have been shown to be subject to a variety of other attacks, such as, adversarial attacks on object detectors [27] or classifiers [13, 29, 30, 32], malicious algorithms which discreetly memorize the training data samples [39], and malicious classifiers which mislead t”
- Mitigating Evasion Attacks to Deep Neural Networks via Region-based Classification (2017, methodology): “DNNs are vulnerable to aacks at testing time [2, 3, 11, 15, 17, 18, 22].”
- Feature Squeezing: Detecting Adversarial Examples in Deep Neural Networks (2017, methodology): “Subsequent papers have explored other strategies to generate adversarial manipulations, including using the linear assumption behind a model [10, 28], saliency maps [34], and evolutionary algorithms [29].”
- On Detecting Adversarial Perturbations (2017, methodology): “Actually, Moosavi-Dezfooli et al. (2016a) show that universal perturbations are not unique and that there even exist many different universal perturbations which have little in common.”