CoLD: Collaborative Label Denoising Framework for Network Intrusion Detection
Shuo Yang
Network and Distributed System Security (NDSS) Symposium 2026 · Day 3 · Web Security
CoLD (Collaborative Label Denoising) is a framework that addresses a fundamental but often overlooked problem in network intrusion detection: **label noise** in training data. The researchers demonstrate that mislabeled data in IDS datasets causes **15-50% performance degradation** at 20-40% noise levels, leading to high false positive rates that burden security teams. Through **causal analysis**, they identify the root cause as **local consistency** -- a phenomenon where features from different traffic categories share similar distributions, amplifying the impact of mislabeled samples.
AI review
A theoretically motivated framework for handling label noise in IDS training data, using causal analysis and self-supervised learning. The causal model identifying local consistency as the root cause is interesting, but the practical impact is modest (5-6% improvement), the presentation was delivered by a stand-in who couldn't address technical details, and the evaluation lacks adversarial testing against real evasion techniques.