Anomaly Detection in Enterprise Information Systems Based on Deep Learning

This study presents an innovative anomaly detection framework for enterprise information systems, integrating deep learning and multi-model ensemble techniques. The framework has been evaluated on a large-scale real-world dataset, demonstrating superior performance. Experimental results show significant achievements in key metrics such as accuracy, F1 score, and AUC value. Compared to traditional methods, this framework exhibits notable improvements in detection precision and processing speed. It excels in identifying critical anomalies such as system crashes and network attacks, effectively reducing false positive and false negative rates. The framework also shows excellent scalability and adaptability, capable of handling new anomalies and dynamically changing system environments. The research findings provide robust support for enhancing the security and reliability of enterprise information systems, offering broad practical application prospects.