Dynamic Epsilon Scheduling: A Multi-Factor Adaptive Perturbation Budget for Adversarial Training

TL;DR

This paper introduces Dynamic Epsilon Scheduling (DES), an adaptive framework that adjusts adversarial perturbation budgets per instance during training, improving robustness and accuracy of neural networks against adversarial attacks.

Contribution

The paper proposes a novel, data-driven method for adaptively tuning perturbation budgets in adversarial training based on multiple instance-specific factors.

Findings

Improves adversarial robustness on CIFAR datasets.

Enhances standard accuracy compared to fixed-epsilon methods.

Provides theoretical analysis of stability and convergence.

Abstract

Adversarial training is among the most effective strategies for defending deep neural networks against adversarial examples. A key limitation of existing adversarial training approaches lies in their reliance on a fixed perturbation budget, which fails to account for instance-specific robustness characteristics. While prior works such as IAAT and MMA introduce instance-level adaptations, they often rely on heuristic or static approximations of data robustness. In this paper, we propose Dynamic Epsilon Scheduling (DES), a novel framework that adaptively adjusts the adversarial perturbation budget per instance and per training iteration. DES integrates three key factors: (1) the distance to the decision boundary approximated via gradient-based proxies, (2) prediction confidence derived from softmax entropy, and (3) model uncertainty estimated via Monte Carlo dropout. By combining these cues into a unified scheduling strategy, DES tailors the perturbation budget dynamically to guide more effective adversarial learning. Experimental results on CIFAR-10 and CIFAR-100 show that our method consistently improves both adversarial robustness and standard accuracy compared to fixed-epsilon baselines and prior adaptive methods. Moreover, we provide theoretical insights into the stability and convergence of our scheduling policy. This work opens a new avenue for instance-aware, data-driven adversarial training methods.