Adversarial examples for models of code
Open accessThe DOI matches its Crossref record; 17 of the 17 references it cites were found in a registry or an index.
Abstract
Neural models of code have shown impressive results when performing tasks such as predicting method names and identifying certain kinds of bugs. We show that these models are vulnerable to adversarial examples , and introduce a novel approach for attacking trained models of code using adversarial examples. The main idea of our approach is to force a given trained model to make an incorrect prediction, as specified by the adversary, by introducing small perturbations that do not change the program’s semantics, thereby creating an adversarial example. To find such perturbations, we present a new technique for Discrete Adversarial Manipulation of Programs (DAMP). DAMP works by deriving the desired prediction with respect to the model’s inputs , while holding the model weights constant, and following the gradients to slightly modify the input code. We show that our DAMP attack is effective across three neural architectures: code2vec, GGNN, and GNN-FiLM, in both Java and C#. Our evaluations demonstrate that DAMP has up to 89% success rate in changing a prediction to the adversary’s choice (a targeted attack) and a success rate of up to 94% in changing a given prediction to any incorrect prediction (a non-targeted attack). To defend a model against such attacks, we empirically examine a variety of possible defenses and discuss their trade-offs. We show that some of these defenses can dramatically drop the success rate of the attacker, with a minor penalty of 2% relative degradation in accuracy when they are not performing under attack. Our code, data, and trained models are available at https://github.com/tech-srl/adversarial-examples .
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
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474
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
Hochreiter, S., & Schmidhuber, J. (1997) · Neural Computation · doi:10.1162/neco.1997.9.8.1735 · Cited by 102,318
The graph neural network model
Scarselli, F., Gori, M., Tsoi, A. C., Hagenbuchner, M., & Monfardini, G. (2009) · IEEE Transactions on Neural Networks · doi:10.1109/tnn.2008.2005605 · Cited by 9,582 · cited 1×
“its inputs, and it represents variable names using a vocabulary of learned embeddings. GatedGraphNeuralNetworks(GGNNs)were introduced by Li et al. [29] as an extension to Graph Neural Networks (GNNs) [45]. The main idea is to represent the problem as a graph, and to aggregate the incoming messages to each vertex with…”
BackgroundConditional Generative Adversarial Nets
Mirza, M., & Osindero, S. (2014) · arXiv · doi:10.48550/arxiv.1411.1784 · Cited by 8,863
Intriguing properties of neural networks
Szegedy, C., Zaremba, W., Sutskever, I., Bruna, J., Erhan, D., Goodfellow, I., & Fergus, R. (2013) · arXiv · doi:10.48550/arxiv.1312.6199 · Cited by 8,143
Recurrent neural network based language model
Mikolov, T., Karafiát, M., Burget, L., Černocký, J., & Khudanpur, S. (2010) · Interspeech 2010 · doi:10.21437/interspeech.2010-343 · Cited by 5,390
Adversarial examples in the physical world
Kurakin, A., Goodfellow, I. J., & Bengio, S. (2018) · Artificial Intelligence Safety and Security · doi:10.1201/9781351251389-8 · Cited by 5,375
DeepFool: A Simple and Accurate Method to Fool Deep Neural Networks
Moosavi-Dezfooli, S.-M., Fawzi, A., & Frossard, P. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.282 · Cited by 5,317 · cited 1×
“Generation of adversarial examples was demonstrated for image classification [Goodfellow et al. 2014a,b; Kurakin et al. 2016; Mirza and Osindero 2014; Moosavi-Dezfooli et al. 2016; Nguyen et al. 2015; Papernot et al. 2017, 2016; Szegedy et al. 2013] and for other domains [Alzantot et al.”
MethodThe Limitations of Deep Learning in Adversarial Settings
Papernot, N., McDaniel, P., Jha, S., Fredrikson, M., Celik, Z. B., & Swami, A. (2016) · 2016 IEEE European Symposium on Security and Privacy (EuroS&P) · doi:10.1109/eurosp.2016.36 · Cited by 3,993
Cited by
Code Prediction by Feeding Trees to Transformers
Kim, S., Zhao, J., Tian, Y., & Chandra, S. (2021) · 2021 IEEE/ACM 43rd International Conference on Software Engineering (ICSE) · doi:10.1109/icse43902.2021.00026 · Cited by 209
Natural Attack for Pre-trained Models of Code
Yang, Z., Shi, J., He, J., & Lo, D. (2022) · Proceedings of the 44th International Conference on Software Engineering · doi:10.1145/3510003.3510146 · Cited by 166
An extensive study on pre-trained models for program understanding and generation
Zeng, Z., Tan, H., Zhang, H., Li, J., Zhang, Y., & Zhang, L. (2022) · Proceedings of the 31st ACM SIGSOFT International Symposium on Software Testing and Analysis · doi:10.1145/3533767.3534390 · Cited by 153
Semantic Robustness of Models of Source Code
Henkel, J., Ramakrishnan, G., Wang, Z., Albarghouthi, A., Jha, S., & Reps, T. (2022) · 2022 IEEE International Conference on Software Analysis, Evolution and Reengineering (SANER) · doi:10.1109/saner53432.2022.00070 · Cited by 92
“[25] present a technique, Discrete Adversarial Manipulation of Programs (DAMP), to explore the adversarial robustness of code models.”
BackgroundMethodInfluentialYou see what I want you to see: poisoning vulnerabilities in neural code search
Wan, Y., Zhang, S., Zhang, H., Sui, Y., Xu, G., Yao, D., Jin, H., & Sun, L. (2022) · Proceedings of the 30th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering · doi:10.1145/3540250.3549153 · Cited by 60
How Effectively Do Code Language Models Understand Poor-Readability Code?
Hu, C., Chai, Y., Zhou, H., Meng, F., Zhou, J., & Gu, X. (2024) · Proceedings of the 39th IEEE/ACM International Conference on Automated Software Engineering · doi:10.1145/3691620.3695072 · Cited by 6
Demonstration Attack against In-Context Learning for Code Intelligence
Ge, Y., Sun, W., Lou, Y., Fang, C., Zhang, Y., Li, Y., Zhang, X., Liu, Y., Zhao, Z., & Chen, Z. (2024) · arXiv · doi:10.48550/arxiv.2410.02841
“Zhang et al. [52] introduced adversarial samples by renaming identifiers through a Metropolis-Hastings sampling-based technique, while Yefet et al. [51] employed gradient-based exploration methods for such attacks.”
Method
Metadata
- Authors
- Yefet, N., Alon, U., & Yahav, E.
- Venue
- Proceedings of the ACM on Programming Languages
- Published
- 2020-11-13
- Volume
- 4
- Issue
- OOPSLA
- Pages
- 1-30
- ISSN
- 2475-1421 (ISSN-L)
- OpenAlex
- W2980634774
- Cited by
- 161
- References
- 17
- FWCI
- 7.10
- Type
- journal article
- Language
- en
- Open access
- diamond
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
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.