LATIDIA · Ciberseguridad

El papel del aprendizaje en el ataque a la detección de intrusiones en la red basada en ML

arXiv: 2602.10299v3Announce Type: replace Resumen: Los sistemas de detección de intrusiones en la red (ML-NIDS) basados en el aprendizaje automático pueden ser eludidos por perturbaciones adversas rudimentarias. El trabajo reciente se ha centrado en identificar quién

La noticia

arXiv:2602.10299v3 Announce Type: replace Abstract: Machine Learning-based Network Intrusion Detection Systems (ML-NIDS) can be bypassed by rudimentary adversarial perturbations. Recent work has focused on identifying where such perturbations can realistically be applied by a host-side adversary. Yet every one of these attacks produces perturbations the same way: searching from scratch for every flow. The cost of an attack therefore grows in lockstep with the number of flows it must perturb, and real networks produce them by the tens of millions. In this paper, we show that using reinforcement learning to train lightweight perturbation-generating policies lets an adversary amortize that cost across flows it perturbs. Counting every detector query and every