mlsanityData to 7 Oct 2026

Spectral Signatures in Backdoor Attacks

arXiv 1811.00636 ↗First posted 1 Nov 2018cs.LG cs.CR stat.MLcited by 457

Not on the map. Why?

Cited by 148 in LLM safety and security, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

A recent line of work has uncovered a new form of data poisoning: so-called backdoor attacks. These attacks are particularly dangerous because they do not affect a network's behavior on typical, benign data. Rather, the network only deviates from its expected output when triggered by a perturbation planted by an adversary. In this paper, we identify a new property of all known backdoor attacks, which we call spectral signatures. This property allows us to utilize tools from robust statistics to thwart the attacks. We demonstrate the efficacy of these signatures in detecting and removing poisoned examples on real image sets and state of the art neural network architectures. We believe that understanding spectral signatures is a crucial first step towards designing ML systems secure against such backdoor attacks

Key facts. Cited by 457 papers in the map (data through 7 Oct 2026). First posted on arXiv 1 Nov 2018.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 57; 2022: 100; 2023: 88; 2024: 93; 2025: 55; 2026: 64.

457 papers.

Sort:

By area:

  1. 2021Neural Attention Distillation: Erasing Backdoor Triggers from Deep Neural Networkscited by 313
  2. 2023Poisoning Web-Scale Training Datasets is Practicalcited by 276
  3. 2021Anti-Backdoor Learning: Training Clean Models on Poisoned Datacited by 258
  4. 2021Adversarial Neuron Pruning Purifies Backdoored Deep Modelscited by 230
  5. 2021Adversarial Unlearning of Backdoors via Implicit Hypergradientcited by 160
  6. 2021Rethinking the Backdoor Attacks' Triggers: A Frequency Perspectivecited by 155
  7. 2022Backdoor Defense via Decoupling the Training Processcited by 149
  8. 2023Dataset Distillation: A Comprehensive Reviewcited by 126
  9. 2022BackdoorBench: A Comprehensive Benchmark of Backdoor Learningcited by 124
  10. 2021SPECTRE: Defending Against Backdoor Attacks Using Robust Statisticscited by 123
  11. 2021BadEncoder: Backdoor Attacks to Pre-trained Encoders in Self-Supervised Learningcited by 120
  12. 2021Sleeper Agent: Scalable Hidden Trigger Backdoors for Neural Networks Trained from Scratchcited by 98
38 more
  1. 2022Truth Serum: Poisoning Machine Learning Models to Reveal Their Secretscited by 94
  2. 2022Wild Patterns Reloaded: A Survey of Machine Learning Security against Training Data Poisoningcited by 93
  3. 2023Prompt as Triggers for Backdoor Attack: Examining the Vulnerability in Language Modelscited by 87
  4. 2024Harmful Fine-tuning Attacks and Defenses for Large Language Models: A Surveycited by 86
  5. 2021Turn the Combination Lock: Learnable Textual Backdoor Attacks via Word Substitutioncited by 85
  6. 2022BppAttack: Stealthy and Efficient Trojan Attacks against Deep Neural Networks via Image Quantization and Contrastive Adversarial Learningcited by 82
  7. 2022Untargeted Backdoor Watermark: Towards Harmless and Stealthy Dataset Copyright Protectioncited by 80
  8. 2022Data-free Backdoor Removal based on Channel Lipschitznesscited by 77
  9. 2024BadChain: Backdoor Chain-of-Thought Prompting for Large Language Modelscited by 74
  10. 2023Unleashing the potential of prompt engineering for large language modelscited by 72
  11. 2021Black-box Detection of Backdoor Attacks with Limited Information and Datacited by 71
  12. 2022A Unified Evaluation of Textual Backdoor Learning: Frameworks and Benchmarkscited by 70
  13. 2021Spinning Language Models: Risks of Propaganda-As-A-Service and Countermeasurescited by 69
  14. 2022The "Beatrix'' Resurrections: Robust Backdoor Detection via Gram Matricescited by 62
  15. 2023CleanCLIP: Mitigating Data Poisoning Attacks in Multimodal Contrastive Learningcited by 62
  16. 2022Backdoor Defense with Machine Unlearningcited by 61
  17. 2021Better Safe Than Sorry: Preventing Delusive Adversaries with Adversarial Trainingcited by 58
  18. 2021Handcrafted Backdoors in Deep Neural Networkscited by 57
  19. 2021T-Miner: A Generative Approach to Defend Against Trojan Attacks on DNN-based Text Classificationcited by 56
  20. 2023Reconstructive Neuron Pruning for Backdoor Defensecited by 56
  21. 2022Planting Undetectable Backdoors in Machine Learning Modelscited by 55
  22. 2022MM-BD: Post-Training Detection of Backdoor Attacks with Arbitrary Backdoor Pattern Types Using a Maximum Margin Statisticcited by 53
  23. 2021AEVA: Black-box Backdoor Detection Using Adversarial Extreme Value Analysiscited by 52
  24. 2021Poison Ink: Robust and Invisible Backdoor Attackcited by 50
  25. 2022Measuring the Effect of Training Data on Deep Learning Predictions via Randomized Experimentscited by 50
  26. 2021Targeted Attack against Deep Neural Networks via Flipping Limited Weight Bitscited by 49
  27. 2023TrojanPuzzle: Covertly Poisoning Code-Suggestion Modelscited by 47
  28. 2023Detecting Backdoors During the Inference Stage Based on Corruption Robustness Consistencycited by 45
  29. 2023UNICORN: A Unified Backdoor Trigger Inversion Frameworkcited by 44
  30. 2023Backdoor Defense via Adaptively Splitting Poisoned Datasetcited by 43
  31. 2024Stress-Testing Capability Elicitation With Password-Locked Modelscited by 42
  32. 2021An Overview of Backdoor Attacks Against Deep Neural Networks and Possible Defencescited by 41
  33. 2021Few-shot Backdoor Defense Using Shapley Estimationcited by 40
  34. 2022Towards A Proactive ML Approach for Detecting Backdoor Poison Samplescited by 40
  35. 2021Trigger Hunting with a Topological Prior for Trojan Detectioncited by 39
  36. 2021CoProtector: Protect Open-Source Code against Unauthorized Training Usage with Data Poisoningcited by 39
  37. 2022Training with More Confidence: Mitigating Injected and Natural Backdoors During Trainingcited by 38
  38. 2023Backdoor Defense via Deconfounded Representation Learningcited by 38

50 most cited of 457 All 457 (JSON)