Making machine learning robust against adversarial inputs
Open accessAbstract
Such inputs distort how machine-learning-based systems are able to function in the world as it is.
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
When the stationarity and neutrality assumptions are violated, algorithms yield unreliable predictions that result in dramatically decreased performances.
“When the stationarity and neutrality assumptions are violated, algorithms yield unreliable predictions that result in dramatically decreased performances [15].”
From A decade of social bot detection (2020) · cites this paper as [15]
The vast majority of machine learning algorithms are designed for operating within environments that are stationary and neutral, if not even benign.
“The vast majority of machine learning algorithms are designed for operating within environments that are stationary and neutral, if not even benign.”
From A decade of social bot detection (2020) · cites this paper as [15]
Adversarial machine learning is a paradigm specifically devised for application in scenarios presenting adversaries motivated in fooling learned models.
“Adversarial machine learning is a paradigm specifically devised for application in those scenarios presenting adversaries motivated in fooling learned models [15].”
From A decade of social bot detection (2020) · cites this paper as [15]
Generative adversarial networks (GANs) are a powerful machine learning framework where two competing deep learning networks are jointly trained in a game-theoretic setting.
“Generative adversarial networks (GANs) are a powerful machine learning framework where two competing deep learning networks are jointly trained in a game-theoretic setting [15].”
From A decade of social bot detection (2020) · cites this paper as [15]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
The citation lists haven’t been fetched yet
References 33 · Cited by 358
Research works printed but not linked: 4
- Dang, H., Yue, H., and Chang, E.C. Evading classifier in the dark: Guiding unpredictable morphing using binary-output blackboxes. arXiv preprint, 2017; https://arxiv.org/pdf/1705.07535.pdf · arXiv
- Goodfellow, I., Bengio, Y., and Courville, A. Deep Learning. MIT Press, Cambridge, MA, 2016; http:// www.deeplearningbook.org/
- Szegedy, C., Zaremba, W., Sutskever, I., Bruna, J., Erhan, D., Goodfellow, I., and Fergus, R. Intriguing properties of neural networks. In Proceedings of the International Conference on Learning Representations, 2014.
- Xu, W., Qi, Y., and Evans, D. Automatically evading classifiers. In Proceedings of the 2016 Network and Distributed Systems Symposium (San Diego, CA, Feb. 21-24). Internet Society, Reston, VA, 2016.
- Authors
- Goodfellow, I., McDaniel, P., & Papernot, N.
- Affiliations
- Google (United States) (US)
- Pennsylvania State University (US)
- Venue
- Communications of the ACM
- Published
- 2018-06-25
- Volume
- 61
- Issue
- 7
- Pages
- 56-66
- ISSN
- 0001-0782 (ISSN-L)
- 1557-7317
- OpenAlex
- W2809895662
- Cited by
- 358
- References
- 33
- Citations by year
- 2026: 18 · 2025: 33 · 2024: 39 · 2023: 42 · 2022: 46 · 2021: 58 · 2020: 70 · 2019: 43 · 2018: 9
- FWCI
- 25.36
- Type
- journal article
- Language
- en
- Open access
- bronze
- doi.org
- Locations
- Communications of the ACM (publishedVersion)
- Funders
- ARL (W911NF-13-2-0045)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Adversarial Robustness in Machine Learning
- Physical Sciences › Computer Science › Artificial Intelligence › Security and Verification in Computing
- Physical Sciences › Computer Science › Artificial Intelligence › Privacy-Preserving Technologies in Data
- Keywords
- adversarial learning, robustness, adversarial examples, machine learning security, attack resistance
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
