English

Ignition Phase : Standard Training for Fast Adversarial Robustness

Machine Learning 2025-10-14 v2 Artificial Intelligence

Abstract

Adversarial Training (AT) is a cornerstone defense, but many variants overlook foundational feature representations by primarily focusing on stronger attack generation. We introduce Adversarial Evolution Training (AET), a simple yet powerful framework that strategically prepends an Empirical Risk Minimization (ERM) phase to conventional AT. We hypothesize this initial ERM phase cultivates a favorable feature manifold, enabling more efficient and effective robustness acquisition. Empirically, AET achieves comparable or superior robustness more rapidly, improves clean accuracy, and cuts training costs by 8-25\%. Its effectiveness is shown across multiple datasets, architectures, and when augmenting established AT methods. Our findings underscore the impact of feature pre-conditioning via standard training for developing more efficient, principled robust defenses. Code is available in the supplementary material.

Keywords

Cite

@article{arxiv.2506.15685,
  title  = {Ignition Phase : Standard Training for Fast Adversarial Robustness},
  author = {Wang Yu-Hang and Liu ying and Fang liang and Wang Xuelin and Junkang Guo and Shiwei Li and Lei Gao and Jian Liu and Wenfei Yin},
  journal= {arXiv preprint arXiv:2506.15685},
  year   = {2025}
}

Comments

Due to errors in both the theoretical formulation and the experimental design, we hereby withdraw the manuscript

R2 v1 2026-07-01T03:24:03.414Z