A5009113689- Network Security and Intrusion Detection
- Anomaly Detection Techniques and Applications
- Complex Network Analysis Techniques
- Advanced Malware Detection Techniques
With the development of Internet technology, open networks and standardized protocols bring more potential security threats. How to propose effective identification prevention methods for some typical malicious program attacks is an important issue be considered active defense. However, most do not analyze risk in attack behavior, so this paper proposes a model based on dependency analysis. The takes character text as input, behavior string subjected word splitting embedding operations....
For attacks on systems during operation, the high false positive rate of traditional machine learning models is no longer able to detect cyber with accuracy. In this paper, we propose a behavior detection model based sequence and graph contrastive learning. The uses instruction strings from UNIX system records as input potential internal attacks. To better learn information, use two parallel algorithms Long Short-Term Memory Network (LSTM) Graph Attention (GAT), encode their complex...