Design of an Automatic Network Anomaly Detection Algorithm Based on Improved Machine Learning
Xinjiu Xie, Jinxian Zhang · International Journal of High Speed Electronics and Systems · 2025
To enhance data anomaly detection in multi-modal topology compressed sensing networks, this paper proposes an automatic detection algorithm based on improved machine learning. Unlike traditional Bayesian audit models that rely on static assumptions, we introduce a residual coupling machine learning framework integrated with inter-layer feature modeling to adaptively reconstruct abnormal data patterns. The method constructs an empirical mode decomposition model for data transmission features, performs dynamic anomaly extraction, and reorganizes sparse feature spaces using a block machine adaptive learning network. Experimental results demonstrate that our approach significantly improves anomaly identification accuracy, reduces network energy expenditure, and prolongs network lifespan compared to conventional techniques, confirming its value for secure and efficient data transmission management in complex network environments.