cs.CVMay 2, 2026

Certified vs. Empirical Adversarial Robust-ness via Hybrid Convolutions with Attention Stochasticity

Authors: Joy DharSong XiaManish Kumar PandeyMaryam HaghighatAzadeh AlaviFerdous SohelWenyu ZhangNayyar Zaidi

Organizations: 1Indian Institute of Technology Ropar · 2Nanyang Technological University · 3RoentGen Health · 4Queensland University of Technology · 5RMIT University · 6Murdoch University · † Independent researcher · 7Deakin University

Abstract

We introduce Hybrid Convolutions with Attention Stochasticity (HyCAS), an adversarial defense that narrows the long-standing gap between provable robustness under L2 certificates and empirical robustness against strong L attacks, while preserving strong generalization across diverse imaging benchmarks. HyCAS unifies deterministic and randomized principles by coupling 1-Lipschitz, spectrally normalized convolutions with two stochastic components, spectral normalized random, projection filters and a randomized attention-noise mechanism, to realize a randomized defense. Injecting smoothing randomness inside the architecture yields an overall <= 2-Lipschitz network with formal certificates. Exten-sive experiments on diverse imaging benchmarks, including CIFAR-10/100, ImageNet-1k, NIH Chest X-ray, HAM10000, show that HyCAS surpasses prior leading certified and empirical defenses, boosting certified accuracy by up to 7.3% (on NIH Chest X-ray) and empirical robustness by up to 3.1% (on HAM10000), without sacrificing clean accuracy. These results show that a randomized Lipschitz constrained architecture can simultaneously improve both certified L2 and empirical L adversarial robustness, thereby supporting safer deployment of deep models in high-stakes applications. Code: https://github.com/misti1203/HyCAS

Explore similar work

CardsList
  1. Improving Certified Robustness via Adversarial Distillation

    Jun 30, 2026Matteo Melis, Jesus Martinez Del Rincon, Vishal SharmaCertificationChain-Of-Thought Distillation