mlsanityData to 7 Oct 2026

The Power of Decaying Steps: Enhancing Attack Stability and Transferability for Sign-based Optimizers

arXiv 2602.19096 ↗First posted 22 Feb 2026cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

Crafting adversarial examples can be formulated as an optimization problem. While sign-based optimizers such as I-FGSM and MI-FGSM have become the de facto standard for the induced optimization problems, there still exist several unsolved problems in theoretical grounding and practical reliability especially in non-convergence and instability, which inevitably influences their transferability. Contrary to the expectation, we observe that the attack success rate may degrade sharply when more number of iterations are conducted. In this paper, we address these issues from an optimization perspective. By reformulating the sign-based optimizer as a specific coordinate-wise gradient descent, we argue that one cause for non-convergence and instability is their non-decaying step-size scheduling. Based upon this viewpoint, we propose a series of new attack algorithms that enforce Monotonically Decreasing Coordinate-wise Step-sizes (MDCS) within sign-based optimizers. Typically, we further provide theoretical guarantees proving that MDCS-MI attains an optimal convergence rate of O(1/√T), where T is the number of iterations. Extensive experiments on image classification and cross-modal retrieval tasks demonstrate that our approach not only significantly improves transferability but also enhances attack stability compared to state-of-the-art sign-based methods.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 37 of its 51 references that are in the map. First posted on arXiv 22 Feb 2026.

Builds on

37 of 51 references.

References · 37

  1. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
  2. 2014Microsoft COCO: Common Objects in Contextcited by 17,498
  3. 2018The Unreasonable Effectiveness of Deep Features as a Perceptual Metriccited by 11,029
  4. 2014Explaining and Harnessing Adversarial Examplescited by 7,287
  5. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,326
  6. 2023Improved Baselines with Visual Instruction Tuningcited by 4,362
  7. 2016Adversarial examples in the physical worldcited by 1,762
  8. 2019Towards VQA Models That Can Readcited by 1,696
  9. 2020Reliable evaluation of adversarial robustness with an ensemble of diverse parameter-free attackscited by 1,442
  10. 2019On the Convergence of Adam and Beyondcited by 834
  11. 2018Improving Transferability of Adversarial Examples with Input Diversitycited by 591
  12. 2019Evading Defenses to Transferable Adversarial Examples by Translation-Invariant Attackscited by 440
25 more
  1. 2021Rethinking Spatial Dimensions of Vision Transformerscited by 361
  2. 2019Nesterov Accelerated Gradient and Scale Invariance for Adversarial Attackscited by 352
  3. 2024Prismatic VLMs: Investigating the Design Space of Visually-Conditioned Language Modelscited by 336
  4. 2019Error Feedback Fixes SignSGD and other Gradient Compression Schemescited by 306
  5. 2021Admix: Enhancing the Transferability of Adversarial Attackscited by 155
  6. 2022Towards Adversarial Attack on Vision-Language Pre-training Modelscited by 142
  7. 2021Visformer: The Vision-friendly Transformercited by 137
  8. 2020IMRAM: Iterative Matching with Recurrent Attention Memory for Cross-Modal Image-Text Retrievalcited by 115
  9. 2019CAMP: Cross-Modal Adaptive Message Passing for Text-Image Retrievalcited by 110
  10. 2023Set-level Guidance Attack: Boosting Adversarial Transferability of Vision-Language Pre-training Modelscited by 106
  11. 2023On the Robustness of Large Multimodal Models Against Image Adversarial Attackscited by 89
  12. 2021Stochastic Variance Reduced Ensemble Adversarial Attack for Boosting the Adversarial Transferabilitycited by 84
  13. 2019Towards Large yet Imperceptible Adversarial Image Perturbations with Perceptual Color Distancecited by 72
  14. 2023Boosting Adversarial Transferability by Block Shuffle and Rotationcited by 64
  15. 2023Structure Invariant Transformation for better Adversarial Transferabilitycited by 61
  16. 2021Boosting Adversarial Transferability through Enhanced Momentumcited by 53
  17. 2017Variants of RMSProp and Adagrad with Logarithmic Regret Boundscited by 50
  18. 2023Boosting Adversarial Transferability by Achieving Flat Local Maximacited by 48
  19. 2023An Adaptive Model Ensemble Adversarial Attack for Boosting Adversarial Transferabilitycited by 42
  20. 2024Boosting Transferability in Vision-Language Attacks via Diversification along the Intersection Region of Adversarial Trajectorycited by 35
  21. 2023Improving Adversarial Transferability via Intermediate-level Perturbation Decaycited by 29
  22. 2024Semantic-Aligned Adversarial Evolution Triangle for High-Transferability Vision-Language Attackcited by 18
  23. 2024VQAttack: Transferable Adversarial Attacks on Visual Question Answering via Pre-trained Modelscited by 10
  24. 2024Improving Integrated Gradient-based Transferable Adversarial Examples by Refining the Integration Pathcited by 10
  25. 2021The Role of Momentum Parameters in the Optimal Convergence of Adaptive Polyak's Heavy-ball Methodscited by 7

Built on by

Not cited yet.