Network Intrusion Detection Based on Deep Belief Network Broad Equalization Learning System

• DOI: 10.3390/electronics13153014
• Publication Date: 2024-07-31T12:02:27Z
• Authors: Miaolei Deng; Chuanchuan Sun; Yupei Kan; Haihang Xu; Xin Zhou; Shaojun Fan
• Keywords: Feature (linguistics); False alarm

Network intrusion detection systems are an important defense technology to guarantee information security and protect a network from attacks. In recent years, the broad learning system has attracted much attention been introduced into with some success. However, since traditional is simple linear structure, when dealing imbalanced datasets, it often ignores feature of minority class samples, leading poorer recognition rate samples. Secondly, high dimensionality redundant features in datasets also seriously affect training time performance system. To address above problems, we propose deep belief equalization The model fully learns large-scale high-dimensional dataset via represents as optimal low-dimensional dataset, then introduces loss v2 reweighing idea classify was experimentally tested using CICIDS2017 validated CICIDS2018 dataset. Compared other algorithms same field, shortens low false alarm rate.