Experiments on Adaptive Techniques for Host-Based Intrusion Detection

摘要: This research explores four experiments of adaptive host-based intrusion detection (ID) techniques in an attempt to develop systems that can detect novel exploits. The technique considered have the most potential is critic designs (ACDs) because their utilization reinforcement learning, which allows learning exploits are difficult pinpoint sensor data. Preliminary results ID using ACD, Elman recurrent neural network, and a statistical anomaly demonstrate ability learn distinguish between clean exploit We used Solaris Basic Security Module (BSM) as data source performed considerable preprocessing on raw A approach called generalized signature-based recommended middle ground ID, has inability exploits, detection, detects too many events including not primary use custom for network-based this application environment.