基于特征选取与树状Parzen估计的入侵检测

针对目前网络空间安全形势快速变化带来的新风险和新挑战, 提出一种基于相关性分析的特征选取和树状Parzen估计优化的入侵检测方法。首先, 通过基于相关性分析的数据特征选取方法对数据维度进行压缩。其次, 对原始数据集进行特征筛选, 生成新的特征子集。最终, 使用序列模型优化算法中的树状Parzen估计算法对随机森林算法进行模型优化。实验结果表明, 相比其他应用机器学习算法的入侵检测方法, 所提方法在提升综合性能的同时拥有更高的检测效率, 有效地提升了入侵检测技术的实用性。

Intrusion detection based on feature selection and tree Parzen estimation

In response to the new risks and challenges bring about by the rapid changes in the current cyberspace security situation, an intrusion detection method optimized by feature selection based on correlation analysis and tree Parzen estimation (TPE) is proposed. Fistly, the data dimensions are compressed by the method of data feature selection based on correlation analysis. Secondly, feature filtering is performed on the original data set, and a new feature subset is generated. Finally, the random forest algorithm is optimized using the TPE algorithm based on sequential model-based global optimization. Experimental results show that the proposed method has higher detection efficiency while improving the overall performance compared with other intrusion detection methods using machine learning algorithms, and effectively improves the practicability of intrusion detection technology.