Cracking Classifiers for Evasion: A Case Study on the Google's Phishing Pages Filter
The DOI matches its Crossref record; 7 of the 7 references it cites were found in a registry or an index.
Abstract
Various classifiers based on the machine learning techniques have been widely used in security applications. Meanwhile, they also became an attack target of adversaries. Many existing studies have paid much attention to the evasion attacks on the online classifiers and discussed defensive methods. However, the security of the classifiers deployed in the client environment has not got the attention it deserves. Besides, earlier studies only concentrated on the experimental classifiers developed for research purposes only. The security of widely-used commercial classifiers still remains unclear. In this paper, we use the Google's phishing pages filter (GPPF), a classifier deployed in the Chrome browser which owns over one billion users, as a case to investigate the security challenges for the client-side classifiers. We present a new attack methodology targeting on client-side classifiers, called classifiers cracking. With the methodology, we successfully cracked the classification model of GPPF and extracted sufficient knowledge can be exploited for evasion attacks, including the classification algorithm, scoring rules and features, etc. Most importantly, we completely reverse engineered 84.8% scoring rules, covering most of high-weighted rules. Based on the cracked information, we performed two kinds of evasion attacks to GPPF, using 100 real phishing pages for the evaluation purpose. The experiments show that all the phishing pages (100%) can be easily manipulated to bypass the detection of GPPF. Our study demonstrates that the existing client-side classifiers are very vulnerable to classifiers cracking attacks.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
ImageNet classification with deep convolutional neural networks
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017) · Communications of the ACM · doi:10.1145/3065386 · Cited by 112,155
Learning methods for generic object recognition with invariance to pose and lighting
LeCun, Y., Huang, F. J., & Bottou, L. (2004) · Proceedings of the 2004 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2004. CVPR 2004. · doi:10.1109/cvpr.2004.1315150 · Cited by 1,324
Huang, L., Joseph, A. D., Nelson, B., Rubinstein, B. I. P., & Tygar, J. D. (2011) · Proceedings of the 4th ACM workshop on Security and artificial intelligence · doi:10.1145/2046684.2046692 · Cited by 1,194 · cited 1×
“According to the taxonomy of attacks against classifiers [9][10][32], the influences of attacks on the classifier are categorized into two types: (1) causative attacks interfere training process with control over the training data to downgrade the performance of the classifier, and (2) exploratory…”
MethodEvasion Attacks against Machine Learning at Test Time
Biggio, B., Corona, I., Maiorca, D., Nelson, B., Šrndić, N., Laskov, P., Giacinto, G., & Roli, F. (2013) · Lecture Notes in Computer Science · doi:10.1007/978-3-642-40994-3_25 · Cited by 890 · cited 3×
“For example, a poisoning attack method against support vector machines (SVM) is presented in [18].”
BackgroundMethodLowd, D., & Meek, C. (2005) · Proceedings of the eleventh ACM SIGKDD international conference on Knowledge discovery in data mining · doi:10.1145/1081870.1081950 · Cited by 755 · cited 2×
“…desirable knowledge [25], perform an adversarial learning to get sufficient knowledge about the target classifier to construct evasion attacks [40], or reconstruct an imitation of the target classifier based on the available public information (e.g., training data) to gain key knowledge [51].”
BackgroundNightmare at test time: robust learning by feature deletion
Globerson, A., & Roweis, S. (2006) · Proceedings of the 23rd international conference on Machine learning - ICML '06 · doi:10.1145/1143844.1143889 · Cited by 309
Stackelberg games for adversarial prediction problems
Brückner, M., & Scheffer, T. (2011) · Proceedings of the 17th ACM SIGKDD international conference on Knowledge discovery and data mining · doi:10.1145/2020408.2020495 · Cited by 229 · cited 2×
“Game-theoretical approaches [22][23] model the interactions between the adversary and the classifier as a game.”
Background
Cited by
Accessorize to a Crime: Real and Stealthy Attacks on State-of-the-Art Face Recognition
Sharif, M., Bhagavatula, S., Bauer, L., & Reiter, M. K. (2016) · Proceedings of the 2016 ACM SIGSAC Conference on Computer and Communications Security · doi:10.1145/2976749.2978392 · Cited by 1,530
Local Model Poisoning Attacks to Byzantine-Robust Federated Learning
Fang, M., Cao, X., Jia, J., & Gong, N. Z. (2019) · arXiv · doi:10.48550/arxiv.1911.11815 · Cited by 828
“Moreover, an attacker could also compromise the confidentiality/intellectual property of a model provider via stealing its model parameters and hyperparameters [34, 58, 60].”
BackgroundThe Role of Machine Learning in Cybersecurity
Apruzzese, G., Laskov, P., Montes de Oca, E., Mallouli, W., Brdalo Rapa, L., Grammatopoulos, A. V., & Di Franco, F. (2023) · Digital Threats: Research and Practice · doi:10.1145/3545574 · Cited by 279
(2024) · doi:10.1109/sp54263.2024.00182 · Cited by 62
Metadata
- Authors
- Liang, B., Su, M., You, W., Shi, W., & Yang, G.
- Venue
- Proceedings of the 25th International Conference on World Wide Web
- Published
- 2016-04-11
- Pages
- 345-356
- OpenAlex
- W2336566325
- Cited by
- 64
- References
- 7
- FWCI
- 2.90
- Type
- conference paper
- Language
- en
- Open access
- closed
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
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Cracking Classifiers for Evasion: A Case Study on the Google's Phishing Pages Filter. https://app.cochat.ai/papers/10.1145/2872427.2883060
- The paper
- Liang, B., Su, M., You, W., Shi, W., & Yang, G. (2016). Cracking Classifiers for Evasion: A Case Study on the Google's Phishing Pages Filter. Proceedings of the 25th International Conference on World Wide Web. https://doi.org/10.1145/2872427.2883060
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.