改进杂交水稻优化算法的入侵检测特征选择OA

To address the issues of low detection efficiency and insufficient accuracy caused by feature redundancy in high-dimensional data for network intrusion detection,an improved hybrid rice optimization(HRO)algorithm was proposed specifically for feature selection.The population initialization was enhanced by employing an improved Circle chaotic map to increase the quality of the initial solutions.A horizontal and vertical crossover strategy was incorporated during the optimization process to augment population diversity.A Gaussian random walk mechanism was introduced to prevent the algorithm from becoming trapped in local optima.To evaluate the effectiveness of this method,experiments were conducted on UCI benchmark datasets and the NSL-KDD network intrusion detection dataset,validated using three classifiers:KNN,DT,and XGBoost.The experimental results demonstrated that on eight UCI datasets,the improved algorithm reduced the total number of original features to approximately 45%,and its classification performance outperformed the original HRO algorithm in all cases.On the NSL-KDD dataset,the enhanced algorithm reduced the original features to about 41%,achieving a maximum classification accuracy of 85.79%with the selected feature subset.