Different from the traditional single databases, there is a big difference between different layers’ data of multi-level database. The differentiation of categorical attributes is small. Traditional database intrusion detection process is simply to consider the point to point data detection between the layers, without considering the similarity between the layers and ignoring the optimization for detected properties of the applied classification between the levels, resulting in lower detection accuracy. In order to avoid the above-mentioned defects of the conventional algorithm, this paper propos an intrusion detection model of multi-layered network by introducing the coarse-to-fine concept. The intrusion feature of computer database is extracted to be used as the basis for intrusion detection of database. The particle swarm distinguish tree is established to make the hierarchical processing for nodes. Through the probability operation of database intrusion detection in different layers, intrusion detection of multi-layer, distributed and large differences database can be achieved. Experimental results show that the use of the intrusion detection algorithm for multi-layer, distributed and large differences database, can increase the security of the database, ensure the safe operation of the database.
Application of OPTICS and ensemble learning for Database Intrusion Detection
