mlsanityData to 7 Oct 2026

Defending Against Repetitive Backdoor Attacks on Semi-supervised Learning through Lens of Rate-Distortion-Perception Trade-off

arXiv 2407.10180 ↗First posted 14 Jul 2024cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

Semi-supervised learning (SSL) has achieved remarkable performance with a small fraction of labeled data by leveraging vast amounts of unlabeled data from the Internet. However, this large pool of untrusted data is extremely vulnerable to data poisoning, leading to potential backdoor attacks. Current backdoor defenses are not yet effective against such a vulnerability in SSL. In this study, we propose a novel method, Unlabeled Data Purification (UPure), to disrupt the association between trigger patterns and target classes by introducing perturbations in the frequency domain. By leveraging the Rate-Distortion-Perception (RDP) trade-off, we further identify the frequency band, where the perturbations are added, and justify this selection. Notably, UPure purifies poisoned unlabeled data without the need of extra clean labeled data. Extensive experiments on four benchmark datasets and five SSL algorithms demonstrate that UPure effectively reduces the attack success rate from 99.78% to 0% while maintaining model accuracy. Code is available here: urlhttps://github.com/chengyi-chris/UPure.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 38 of its 60 references that are in the map. First posted on arXiv 14 Jul 2024.

Builds on

38 of 60 references.

References · 38

  1. 2016Wide Residual Networkscited by 3,057
  2. 2020FixMatch: Simplifying Semi-Supervised Learning with Consistency and Confidencecited by 2,080
  3. 2017Improved Regularization of Convolutional Neural Networks with Cutoutcited by 1,623
  4. 2019MixMatch: A Holistic Approach to Semi-Supervised Learningcited by 1,390
  5. 2019Unsupervised Data Augmentation for Consistency Trainingcited by 1,057
  6. 2016Temporal Ensembling for Semi-Supervised Learningcited by 1,054
  7. 2017The Perception-Distortion Tradeoffcited by 582
  8. 2018Fine-Pruning: Defending Against Backdooring Attacks on Deep Neural Networkscited by 550
  9. 2021FlexMatch: Boosting Semi-Supervised Learning with Curriculum Pseudo Labelingcited by 537
  10. 2018Spectral Signatures in Backdoor Attackscited by 457
  11. 2016Regularization With Stochastic Transformations and Perturbations for Deep Semi-Supervised Learningcited by 441
  12. 2019STRIP: A Defence Against Trojan Attacks on Deep Neural Networkscited by 437
26 more
  1. 2018Detecting Backdoor Attacks on Deep Neural Networks by Activation Clusteringcited by 429
  2. 2019Hidden Trigger Backdoor Attackscited by 359
  3. 2019ReMixMatch: Semi-Supervised Learning with Distribution Alignment and Augmentation Anchoringcited by 349
  4. 2021Neural Attention Distillation: Erasing Backdoor Triggers from Deep Neural Networkscited by 313
  5. 2019Label-Consistent Backdoor Attackscited by 285
  6. 2023Poisoning Web-Scale Training Datasets is Practicalcited by 276
  7. 2021Anti-Backdoor Learning: Training Clean Models on Poisoned Datacited by 258
  8. 2019Rethinking Lossy Compression: The Rate-Distortion-Perception Tradeoffcited by 207
  9. 2017Towards Poisoning of Deep Learning Algorithms with Back-gradient Optimizationcited by 188
  10. 2017Learning from Noisy Labels with Distillationcited by 160
  11. 2021Adversarial Unlearning of Backdoors via Implicit Hypergradientcited by 160
  12. 2021Rethinking the Backdoor Attacks' Triggers: A Frequency Perspectivecited by 155
  13. 2022Backdoor Defense via Decoupling the Training Processcited by 149
  14. 2021SPECTRE: Defending Against Backdoor Attacks Using Robust Statisticscited by 123
  15. 2021Back to the Drawing Board: A Critical Evaluation of Poisoning Attacks on Production Federated Learningcited by 116
  16. 2022USB: A Unified Semi-supervised Learning Benchmark for Classificationcited by 95
  17. 2021Backdoor Attacks on Self-Supervised Learningcited by 89
  18. 2020DeepSweep: An Evaluation Framework for Mitigating DNN Backdoor Attacks using Data Augmentationcited by 75
  19. 2021Universal Rate-Distortion-Perception Representations for Lossy Compressioncited by 46
  20. 2023Detecting Backdoors During the Inference Stage Based on Corruption Robustness Consistencycited by 45
  21. 2021Poisoning the Unlabeled Dataset of Semi-Supervised Learningcited by 44
  22. 2023Black-box Backdoor Defense via Zero-shot Image Purificationcited by 44
  23. 2022Towards A Proactive ML Approach for Detecting Backdoor Poison Samplescited by 40
  24. 2023Backdoor Defense via Deconfounded Representation Learningcited by 38
  25. 2022An Embarrassingly Simple Backdoor Attack on Self-supervised Learningcited by 30
  26. 2022The Perils of Learning From Unlabeled Data: Backdoor Attacks on Semi-supervised Learningcited by 9

Built on by

Not cited yet.