Research on network intrusion detection model based on multi-granularity cascaded forest optimization algorithm
To address the ever-evolving and diverse nature of large-scale network intrusions and the subsequent cybersecurity threats,this paper proposes a network intrusion detection model based on the Multi-Granularity Cascaded Forest(GCForest).The model initially preprocesses raw data,subsequently incorporates the Fisher Score algorithm to rank different feature information by their weights,and ultimately feeds the ranked feature information into the convolutional layer and forest layer of the cascaded forest for deep feature expression and classification,thereby achieving precise classification results.Validation using the KDD 99 dataset demonstrates that under three experimental scenarios with training set proportions of 90％,70％,and 30％,the model achieves classification accuracies of 98.20％,99.00％,and 99.27％respectively.The experimental results prove that the proposed algo-rithm in this paper can accurately identify various network attacks,providing an effective basis for distinguishing and detecting network intrusions in existing systems.
万方数据
维普
