MAEL: meta-active semi-supervised ensemble learning model for DDoS attack detection

MAEL: meta-active semi-supervised ensemble learning model for DDoS attack detection

연구 분야: Safety

학회: Cluster Computing

In recent years, Distributed Denial of Service (DDoS) attacks have significant harmful consequences on network infrastructure, including the depletion of computational resources and the saturation of communication channels. Today, developing efficient approaches to identify and stop DDoS attacks is with paramount importance in order to protect large networks. However, traditional supervised learning models require large labeled datasets, which are often expensive and time-consuming to produce. To address this, we propose a novel meta-active ensemble learning pipeline that integrates graph-based clustering-driven auto-labeling, active learning and adaptive meta-tuning of classification models. The system dynamically adjusts the classification model using a meta-learner, selecting the most informative samples for labeling while leveraging graph-based labeling for automated annotation. This hybrid approach reduces labeling costs and improves model performance over time. The experimental results demonstrate the effectiveness of the proposed approach, showcasing notable improvements in key performance metrics such as accuracy, precision, and F1-Score.