Skip to content
Log in

Hybrid Hardening for Robust DNNs Under Adversarial Attacks

L. D. de Melo M. Gani Alberto Bosio O. Stan Vlad-Cristian Miclea L. Miclea Rodrigo Possamai Bastos David Novo Bastien Deveautour
Jul 2026 · IEEE International Symposium on On-Line Testing and Robust System Design · pp. 1-5 · 0 citations · 34 references

Abstract

Deep Neural Networks are vulnerable to adversarial perturbations, which can cause failures in safety-critical systems. State-of-the-art countermeasures often incur prohibitive overheads, creating a need for cost-efficient robustness strategies. This paper proposes a hybrid hardening technique integrating Triple Modular Redundancy with Adversarial Training, Label Smoothing, and Iterative Pruning to reduce hardening cost while preserving accuracy. Experimental results demonstrate that our approach maintains up to 86.7% accuracy under strong adversarial perturbations (ϵ = 0.1), outperforming unprotected models by a factor of 4.8x while reducing memory footprint up to 76% through pruning.

DOI Semantic Scholar
View source

Similar papers

Open access Aug 2026

Adversarially Robust Hardware Trojan Detection with Synthetic Data Augmentation

This paper evaluates the robustness of the Support Vector Machine (SVM) classifier, a leading algorithm in state-of-the-art HT detection frameworks, under gradient-based adversarial attacks, and highlights the need to reframe hardware security evaluations beyond nominal accuracy toward adversarial robustness.

Ashutosh Ghimire, Lingwei Chen, Cole Castronova et al. · 0 citations
Preprint Aug 2026

Adversarial Training of Linear Models under Stealthy Attacks

Predictive models are widely used in many fields, but are vulnerable to false data injection attacks. To address this, detection schemes and adversarial training have been proposed, but such approaches lack guarantees against stealthy attacks. We therefore propose a detector-based switched model, in which optimal attack strategies are stealthy. For linear prediction models, we derive a convex formulation of the resulting adversarial risk. The model incorporates protected features and introduces a hyperparameter modelling attack probability, enabling an explicit performance trade-off between clean and attacked data regimes. Numerical simulations on real and synthetic data show improved performance on partially attacked data, even for misspecified attack probabilities.

Lovisa Eriksson, Dave Zachariah, A. Teixeira · 0 citations
Preprint Aug 2026

Understanding Fault Tolerance of Adversarially Robust Pruned Models

Results show that adversarial training improves robustness against input perturbations but increases sensitivity to stuck-at-zero weight faults, highlighting the need to jointly consider adversarial robustness and hardware reliability.

Manali Dangarikar, Cory E. Merkel · 0 citations
Preprint Aug 2026

MOAT: Model-Agnostic Randomized Transformations for preventing Efficiency Degradation Attacks on ViTs

To adopt the Vision Transformers (ViTs) in resource-constrained environment, token pruning is widely used to reduce computational cost without impacting accuracy. However, adversaries have developed targeted attacks against said token pruning techniques to undermine such attempts to make ViTs efficient. In this paper, we propose MOAT, a model-agnostic pre-processing defense pipeline that applies a combination of input transformations to protect efficient ViT implementations against adversarial efficiency attacks. MOAT operates directly on the input without requiring modifications to the model architecture or token pruning mechanism. Experimental results demonstrate that, across all evaluated ViT models, MOAT limits GFLOPs degradation under adversarial attacks to within 3.4% of the original unattacked model.

Anadi Goyal, Nandish Chattopadhyay, C. Karfa et al. · 0 citations

Generalization Analysis for Adversarial Vision Transformers

This work establishes a high-probability generalization bound for ViTs in classification tasks under adversarial settings, and elucidates the roles of several factors in mitigating perturbation effects, norm regularization of weight matrices and depth-wise propagation constraints on layer-wise norms.

Ziwen Jiang, Chang Cao, Han Li et al. · 0 citations
Open access Jul 2026

Needle in a Haystack: Decamouflaging Adversarial Examples Using SBERT Embeddings

This paper introduces a new anagram based attack into the literature and trains the model to learn these representations by adding a novel label to the dataset which is used as a feature for training the model.

Sai Reethi, Sudha Pydi, Pelluri · 0 citations