Adversarial Robustness of Neural-Statistical Features in Detection of Generative Transformers
Open accessThe DOI matches its Crossref record; 18 of the 18 references it cites were found in a registry or an index.
Abstract
The detection of computer-generated text is an area of rapidly increasing significance as nascent generative models allow for efficient creation of compelling human-like text, which may be abused for the purposes of spam, disinformation, phishing, or online influence campaigns. Past work has studied detection of current state-of-the-art models, but despite a developing threat landscape, there has been minimal analysis of the robustness of detection methods to adversarial attacks. To this end, we evaluate neural and non-neural approaches on their ability to detect computer-generated text, their robustness against text adversarial attacks, and the impact that successful adversarial attacks have on human judgement of text quality. We find that while statistical features underperform neural features, statistical features provide additional adversarial robustness that can be leveraged in ensemble detection models. In the process, we find that previously effective complex phrasal features for detection of computer-generated text hold little predictive power against contemporary generative models, and identify promising statistical features to use instead. Finally, we pioneer the usage of $Δ$MAUVE as a proxy measure for human judgement of adversarial text quality.
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
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2018) · arXiv · doi:10.48550/arxiv.1810.04805 · Cited by 73,173
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019) · Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) · doi:10.18653/v1/n19-1423 · Cited by 54,325
Language Models are Few-Shot Learners
Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al. (2020) · arXiv · doi:10.48550/arxiv.2005.14165 · Cited by 23,372
DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter
Sanh, V., Debut, L., Chaumond, J., & Wolf, T. (2019) · arXiv · doi:10.48550/arxiv.1910.01108 · Cited by 4,653
Practical Black-Box Attacks against Machine Learning
Papernot, N., McDaniel, P., Goodfellow, I., Jha, S., Celik, Z. B., & Swami, A. (2017) · Proceedings of the 2017 ACM on Asia Conference on Computer and Communications Security · doi:10.1145/3052973.3053009 · Cited by 3,495 · cited 1×
“rithms leverage machine learning [31] [32] [33].”
CTRL: A Conditional Transformer Language Model for Controllable Generation
Keskar, N. S., McCann, B., Varshney, L. R., Xiong, C., & Socher, R. (2019) · arXiv · doi:10.48550/arxiv.1909.05858 · Cited by 751
MiniLM: Deep Self-Attention Distillation for Task-Agnostic Compression of Pre-Trained Transformers
Wang, W., Wei, F., Dong, L., Bao, H., Yang, N., & Zhou, M. (2020) · arXiv · doi:10.48550/arxiv.2002.10957 · Cited by 703 · cited 2×
“5343 1 768 MiniLM [48] 215M QA Pairs 0.”
Black-Box Generation of Adversarial Text Sequences to Evade Deep Learning Classifiers
Gao, J., Lanchantin, J., Soffa, M. L., & Qi, Y. (2018) · 2018 IEEE Security and Privacy Workshops (SPW) · doi:10.1109/spw.2018.00016 · Cited by 693
Defending Against Neural Fake News
Zellers, R., Holtzman, A., Rashkin, H., Bisk, Y., Farhadi, A., Roesner, F., & Choi, Y. (2019) · arXiv · doi:10.48550/arxiv.1905.12616 · Cited by 569 · cited 2×
“5 billion parameter GPT-2 model (and other implementations of generative pre-trained Transformers with similar scale, such as Grover [4]) have already opened a number of new attack possibilities.”
BackgroundStanza: A Python Natural Language Processing Toolkit for Many Human Languages
Qi, P., Zhang, Y., Zhang, Y., Bolton, J., & Manning, C. D. (2020) · arXiv · doi:10.48550/arxiv.2003.07082 · Cited by 476 · cited 2×
“We first tokenize the input using Stanza [54], then lemmatize the results.”
Method
Cited by
A Survey on LLM-Generated Text Detection: Necessity, Methods, and Future Directions
Wu, J., Yang, S., Zhan, R., Yuan, Y., Wong, D. F., & Chao, L. S. (2023) · arXiv · doi:10.48550/arxiv.2310.14724 · Cited by 212
“…attacks, including cutoff (cropping a portion of the feature or input) [153], shuffle (randomly disrupting the word order of the input) [154], mutation (character and word mutation) [155], word swapping (substituting other suitable words given the context) [156, 157, 158] and misspelling [159].”
BackgroundMachine Generated Text: A Comprehensive Survey of Threat Models and Detection Methods
Crothers, E., Japkowicz, N., & Viktor, H. (2022) · arXiv · doi:10.48550/arxiv.2210.07321 · Cited by 196
“Contemporarymodels for this purpose are neural in nature, and as such, modern feature-based approaches leveraging linguistic features from auxiliary models perform inference on a neural network as part of the creation of feature vectors [40, 60].”
InfluentialSelf-citationMachine-Generated Text: A Comprehensive Survey of Threat Models and Detection Methods
Crothers, E. N., Japkowicz, N., & Viktor, H. L. (2023) · IEEE Access · doi:10.1109/access.2023.3294090 · Cited by 196
“Contemporarymodels for this purpose are neural in nature, and as such, modern feature-based approaches leveraging linguistic features from auxiliary models perform inference on a neural network as part of the creation of feature vectors [40, 60].”
InfluentialSelf-citationA Survey on LLM-Generated Text Detection: Necessity, Methods, and Future Directions
Wu, J., Yang, S., Zhan, R., Yuan, Y., Chao, L. S., & Wong, D. F. (2025) · Computational Linguistics · doi:10.1162/coli_a_00549 · Cited by 212
The Science of Detecting LLM-Generated Text
Tang, R., Chuang, Y.-N., & Hu, X. (2024) · Communications of the ACM · doi:10.1145/3624725 · Cited by 125
“In addition, studies have also shown that using pre-trained language models to extract semantic textual features, followed by SVM for classification, can outperform the use of statistical features alone [7].”
Background
Metadata
- Authors
- Crothers, E., Japkowicz, N., Viktor, H., & Branco, P.
- Venue
- 2022 International Joint Conference on Neural Networks (IJCNN)
- Publisher
- IEEE
- Published
- 2022-07-18
- Pages
- 1-8
- ISSN
- 2161-4407 (ISSN-L)
- OpenAlex
- W4226471298
- Cited by
- 22
- References
- 18
- FWCI
- 3.15
- Type
- conference paper
- Language
- en
- Open access
- green
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: Adversarial Robustness of Neural-Statistical Features in Detection of Generative Transformers. https://app.cochat.ai/papers/10.1109/ijcnn55064.2022.9892269
- The paper
- Crothers, E., Japkowicz, N., Viktor, H., & Branco, P. (2022). Adversarial Robustness of Neural-Statistical Features in Detection of Generative Transformers. 2022 International Joint Conference on Neural Networks (IJCNN). https://doi.org/10.1109/ijcnn55064.2022.9892269
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.