A Heuristic Genetic Neural Network for Intrusion Detection

A Heuristic Genetic Neural Network for Intrusion Detection,10.1109/ICICIS.2011.133,Biying Zhang

A Heuristic Genetic Neural Network for Intrusion Detection  
BibTex | RIS | RefWorks Download
In order to model normal behaviors accurately and improve the performance of intrusion detection, a heuristic genetic neural network(HGNN) is presented. Feature selection, structure design and weight adaptation are evolved jointly in consideration of the interdependence of input features, network structure and connection weights. The penalty factors for the number of input nodes and hidden nodes are introduced into fitness function. The crossover operator based on generated subnet is adopted considering the relationship between genotype and phenotype. An adaptive mutation rate is applied, and the mutation type is selected heuristically from weight adaptation, node deletion and node addition. When the population is not evolved continuously for many generations, in order to jump from the local optima and extend the search space, the mutation rate will be increased and the mutation type will be changed. Experimental results with the KDD-99 dataset show that the HGNN achieves better detection performance in terms of detection rate and false positive rate.
Published in 2011.
Cumulative Annual
View Publication
The following links allow you to view full publications. These links are maintained by other sources not affiliated with Microsoft Academic Search.