LDAS&ET-AD: Learnable Distillation Attack Strategies and Evolvable Teachers Adversarial Distillation
Source abstract
Abstract
Adversarial distillation (AD) has emerged as a potential solution to tackle the challenging optimization problem of loss with hard labels in adversarial training. However, fixed sample-agnostic and student-egocentric attack s... | Find, read and cite all the research you need on Tech Science Press
Source: openalex_inverted_index · Quality: baseline
Research context
OpenAlex topics: Adversarial Robustness in Machine Learning · Advanced Malware Detection Techniques · Network Security and Intrusion Detection
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
References (45)
Show source-provided references
1. Deep convolutional neural network for brain tumor segmentation10.1007/s42835-023-01479-y ↗
2. Survey of hallucination in natural language generation10.1145/3571730 ↗
3. A cross-entropy-guided measure (CEGM) for assessing speech recognition performance and optimizing DNN-based speech enhancement10.1109/TASLP.2020.3036783 ↗
4. Intriguing properties of neural networks
5. Multi-source adversarial sample attack on autonomous vehicles10.1109/TVT.2021.3061065 ↗
6. Effective and robust physical-world attacks on deep learning face recognition systems10.1109/TIFS.2021.3102492 ↗
7. Detect and defense against adversarial examples in deep learning using natural scene statistics and adaptive denoising10.1007/s00521-021-06330-x ↗
8. Chained dual-generative adversarial network: A generalized defense against adversarial attacks
9. ComDefend: An efficient image compression model to defend adversarial examples
10. Theoretically principled trade-off between robustness and accuracy
11. Adversarial training against adversarial attacks for machine learning-based intrusion detection systems
12. LAS-AT: Adversarial training with learnable attack strategy
13. Towards deep learning models resistant to adversarial attacks
14. Certified adversarial robustness within multiple perturbation bounds
15. Boosting verified training for robust image classifications via abstraction
16. DiffSmooth: Certifiably robust learning via diffusion models and local smoothing
17. Improving l1-certified robustness via randomized smoothing by leveraging box constraints
18. On the loss landscape of adversarial training: Identifying challenges and how to overcome them
19. Adversarially robust distillation
20. J. Maroto, G. Ortiz-Jiménez, and P. Frossard, “On the benefits of knowledge distillation for adversarial robustness,” arXiv preprint arXiv:2203.07159, 2022.
21. Revisiting adversarial robustness distillation: Robust soft labels make student better
22. G. Cao et al., “Vanilla feature distillation for improving the accuracy-robustness trade-off in adversarial training,” arXiv preprint arXiv:2206.02158, 2022.
23. Robust overfitting may be mitigated by properly learned smoothening
24. Enhanced accuracy and robustness via multi-teacher adversarial distillation
25. Reliable adversarial distillation with unreliable teachers
26. Learning to teach
27. Deep mutual learning
28. Knowledge distillation by On-the-Fly native ensemble
29. Embedded mutual learning: A novel online distillation method integrating diverse knowledge sources10.1007/s10489-022-03974-7 ↗
30. Switchable online knowledge distillation
31. BERT learns to teach: Knowledge distillation with meta learning
32. Learning knowledge representation with meta knowledge distillation for single image super-resolution10.1016/j.jvcir.2023.103874 ↗
33. Meta pseudo labels
34. Meta pseudo labels for chest x-ray image classification
35. Tailoring instructions to studentss learning levels boosts knowledge distillation
36. Curriculum adversarial training
37. Attacks which do not kill training make adversarial learning stronger
38. On the convergence and robustness of adversarial training
39. Explaining and harnessing adversarial examples
40. Towards evaluating the robustness of neural networks
41. Reliable evaluation of adversarial robustness with an ensemble of diverse parameter-free attacks
42. Improving adversarial robustness requires revisiting misclassified examples
43. A. Krizhevsky, “Learning multiple layers of features from tiny images,” M.S. dissertation, Univ. of Toronto, Canada, 2009.
44. Deep residual learning for image recognition
45. MobileNetV2: Inverted residuals and linear bottlenecks