mlsanityData to 14 Oct 2026

"That Is a Suspicious Reaction!": Interpreting Logits Variation to Detect NLP Adversarial Attacks

arXiv 2204.04636 ↗First posted 10 Apr 2022cs.AI cs.CL cs.LGcited by 24

Not on the map. Why?

Abstract (arXiv)

Adversarial attacks are a major challenge faced by current machine learning research. These purposely crafted inputs fool even the most advanced models, precluding their deployment in safety-critical applications. Extensive research in computer vision has been carried to develop reliable defense strategies. However, the same issue remains less explored in natural language processing. Our work presents a model-agnostic detector of adversarial text examples. The approach identifies patterns in the logits of the target classifier when perturbing the input text. The proposed detector improves the current state-of-the-art performance in recognizing adversarial inputs and exhibits strong generalization capabilities across different NLP models, datasets, and word-level attacks.

Key facts. Cited by 24 papers in the map (data through 14 Oct 2026). Builds on 29 of its 51 references that are in the map. First posted on arXiv 10 Apr 2022.

Builds on

29 of 51 references.

References · 29

  1. 2017Attention Is All You Needcited by 60,361
  2. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
  3. 2014Explaining and Harnessing Adversarial Examplescited by 7,364
  4. 2017A Unified Approach to Interpreting Model Predictionscited by 6,555
  5. 2016XGBoost: A Scalable Tree Boosting Systemcited by 5,166
  6. 2019DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lightercited by 4,307
  7. 2015Character-level Convolutional Networks for Text Classificationcited by 2,330
  8. 2016Adversarial examples in the physical worldcited by 1,769
  9. 2019Is BERT Really Robust? A Strong Baseline for Natural Language Attack on Text Classification and Entailmentcited by 671
  10. 2020On Adaptive Attacks to Adversarial Example Defensescited by 555
  11. 2020BERT-ATTACK: Adversarial Attack Against BERT Using BERTcited by 491
  12. 2018Generating Natural Language Adversarial Examplescited by 424
17 more
  1. 2018Black-box Generation of Adversarial Text Sequences to Evade Deep Learning Classifierscited by 345
  2. 2020BAE: BERT-based Adversarial Examples for Text Classificationcited by 316
  3. 2018Adversarial Example Generation with Syntactically Controlled Paraphrase Networkscited by 286
  4. 2018Adversarial Logit Pairingcited by 224
  5. 2019Certified Robustness to Adversarial Word Substitutionscited by 168
  6. 2019Combating Adversarial Misspellings with Robust Word Recognitioncited by 163
  7. 2019Text Processing Like Humans Do: Visually Attacking and Shielding NLP Systemscited by 101
  8. 2019The Odds are Odd: A Statistical Test for Detecting Adversarial Examplescited by 75
  9. 2021Towards Robustness Against Natural Language Word Substitutionscited by 69
  10. 2019Learning to Discriminate Perturbations for Blocking Adversarial Attacks in Text Classificationcited by 57
  11. 2020Frequency-Guided Word Substitutions for Detecting Textual Adversarial Examplescited by 54
  12. 2019When Explainability Meets Adversarial Learning: Detecting Adversarial Examples using SHAP Signaturescited by 40
  13. 2018Attacks Meet Interpretability: Attribute-steered Detection of Adversarial Samplescited by 38
  14. 2016Early Methods for Detecting Adversarial Imagescited by 36
  15. 2020Defense against Adversarial Attacks in NLP via Dirichlet Neighborhood Ensemblecited by 28
  16. 2019Detecting Adversarial Examples and Other Misclassifications in Neural Networks by Introspectioncited by 16
  17. 2020Detection Defense Against Adversarial Attacks with Saliency Mapcited by 10

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2022: 4; 2023: 8; 2024: 4; 2025: 6; 2026: 2.

24 papers.

Sort:

By area:

  1. 2022State-of-the-art generalisation research in NLP: A taxonomy and reviewcited by 98
  2. 2024Navigating the Risks: A Survey of Security, Privacy, and Ethics Threats in LLM-Based Agentscited by 65
  3. 2024Text Generation: A Systematic Literature Review of Tasks, Evaluation, and Challengescited by 18
  4. 2023Why do universal adversarial attacks work on large language models?: Geometry might be the answercited by 14
  5. 2025Beyond Next Token Probabilities: Learnable, Fast Detection of Hallucinations and Data Contamination on LLM Output Distributionscited by 12
  6. 2023TextShield: Beyond Successfully Detecting Adversarial Sentences in Text Classificationcited by 8
  7. 2023IFAN: An Explainability-Focused Interaction Framework for Humans and NLP Modelscited by 8
  8. 2023Masked Language Model Based Textual Adversarial Example Detectioncited by 5
  9. 2024ROIC-DM: Robust Text Inference and Classification via Diffusion Modelcited by 5
  10. 2022Exploring Adversarial Attacks and Defenses in Vision Transformers trained with DINOcited by 4
  11. 2022An Interpretability Evaluation Benchmark for Pre-trained Language Modelscited by 4
  12. 2025Exploring Gradient-Guided Masked Language Model to Detect Textual Adversarial Attackscited by 4
12 more
  1. 2024SemRoDe: Macro Adversarial Training to Learn Representations That are Robust to Word-Level Attackscited by 3
  2. 2022ADDMU: Detection of Far-Boundary Adversarial Examples with Data and Model Uncertainty Estimationcited by 2
  3. 2023On the Universal Adversarial Perturbations for Efficient Data-free Adversarial Detectioncited by 2
  4. 2025Guided Perturbation Sensitivity (GPS): Detecting Adversarial Text via Embedding Stability and Word Importancecited by 2
  5. 2023Alert-ME: An Explainability-Driven Defense Against Adversarial Examples in Transformer-Based Text Classificationcited by 1
  6. 2023DA3: A Distribution-Aware Adversarial Attack against Language Modelscited by 1
  7. 2025Q-FAKER: Query-free Hard Black-box Attack via Controlled Generationcited by 1
  8. 2023VoteTRANS: Detecting Adversarial Text without Training by Voting on Hard Labels of Transformationscited by 0
  9. 2025RedHerring Attack: Testing the Reliability of Attack Detectioncited by 0
  10. 2025FeatureLens: A Highly Generalizable and Interpretable Framework for Detecting Adversarial Examples Based on Image Featurescited by 0
  11. 2026Rectifying Adversarial Examples Using Their Vulnerabilitiescited by 0
  12. 2026RTD-Guard: A Black-Box Textual Adversarial Detection Framework via Replacement Token Detectioncited by 0