Deep Asymmetric Convolutional Encoder for Enhanced Network Intrusion Detection

Abstract

The rise of Internet technology has led to an increase in cyber threats such as DoS attacks and ransomware. Leveraging advancements in big data and deep learning, this paper introduces a Deep Asymmetric Convolutional Encoder (DACAE) for network intrusion detection. By combining traditional self-encoders with convolutional self-encoders, the DACAE model effectively extracts local optimal features and generates new abstract features. These features are then classified using a random forest algorithm. Experimental results show that the DACAE-RF model outperforms traditional models, especially in detecting small sample data and maintaining high detection levels in multi-sample scenarios, proving its effectiveness in handling diverse and high-dimensional data flows.