mlsanityData to 14 Oct 2026

Two Heads are Actually Better than One: Towards Better Adversarial Robustness via Transduction and Rejection

arXiv 2305.17528 ↗First posted 27 May 2023cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

Both transduction and rejection have emerged as important techniques for defending against adversarial perturbations. A recent work by Goldwasser et al. showed that rejection combined with transduction can give provable guarantees (for certain problems) that cannot be achieved otherwise. Nevertheless, under recent strong adversarial attacks, their work was shown to have low performance in a practical deep-learning setting. In this paper, we take a step towards realizing the promise of transduction+rejection in more realistic scenarios. Our key observation is that a novel application of a reduction technique by Tram`er, which was until now only used to demonstrate the vulnerability of certain defenses, can be used to actually construct effective defenses. Theoretically, we show that a careful application of this technique in the transductive setting can give significantly improved sample-complexity for robust generalization. Our theory guides us to design a new transductive algorithm for learning a selective model; extensive experiments using state of the art attacks show that our approach provides significantly better robust accuracy (81.6% on CIFAR-10 and 57.9% on CIFAR-100 under l_∞ with budget 8/255) than existing techniques.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 25 of its 38 references that are in the map. First posted on arXiv 27 May 2023.

Builds on

25 of 38 references.

References · 25

  1. 2014Explaining and Harnessing Adversarial Examplescited by 7,364
  2. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,415
  3. 2015Domain-Adversarial Training of Neural Networkscited by 3,044
  4. 2016Towards Evaluating the Robustness of Neural Networkscited by 2,998
  5. 2019Theoretically Principled Trade-off between Robustness and Accuracycited by 1,517
  6. 2020Reliable evaluation of adversarial robustness with an ensemble of diverse parameter-free attackscited by 1,451
  7. 2020RobustBench: a standardized adversarial robustness benchmarkcited by 661
  8. 2020On Adaptive Attacks to Adversarial Example Defensescited by 555
  9. 2017Nearly-tight VC-dimension and pseudodimension bounds for piecewise linear neural networkscited by 311
  10. 2019VC Classes are Adversarially Robustly Learnable, but Only Improperlycited by 83
  11. 2021Detecting Adversarial Examples Is (Nearly) As Hard As Classifying Themcited by 48
  12. 2015Sample compression schemes for VC classescited by 47
13 more
  1. 2020Beyond Perturbations: Learning Guarantees with Arbitrary Adversarial Test Examplescited by 44
  2. 2019Deep Neural Rejection against Adversarial Examplescited by 32
  3. 2021Fighting Gradients with Gradients: Dynamic Defenses against Adversarial Attackscited by 20
  4. 2018Sample Compression for Real-Valued Learnerscited by 19
  5. 2019A Research Agenda: Dynamic Models to Defend Against Correlated Attackscited by 18
  6. 2021Two Coupled Rejection Metrics Can Tell Adversarial Examples Apartcited by 14
  7. 2022Increasing Confidence in Adversarial Robustness Evaluationscited by 12
  8. 2021Transductive Robust Learning Guaranteescited by 10
  9. 2019Playing it Safe: Adversarial Robustness with an Abstain Optioncited by 9
  10. 2020ATRO: Adversarial Training with a Rejection Optioncited by 8
  11. 2021Towards Evaluating the Robustness of Neural Networks Learned by Transductioncited by 8
  12. 2022Be Your Own Neighborhood: Detecting Adversarial Example by the Neighborhood Relations Built on Self-Supervised Learningcited by 5
  13. 2022Improving Adversarial Robustness via Joint Classification and Multiple Explicit Detection Classescited by 4

Built on by

Not cited yet.