微型电脑应用2025，Vol.41Issue(2)：9-12,4.
基于异常检测与优化深度学习的入侵检测系统
Intrusion Detection System Based on Anomaly Detection and Optimized Deep Learning
摘要
Abstract
A hybrid intrusion detection system based on anomaly detection and optimized deep learning is proposed for the secur-ity requirements of intrusion detection systems in a pervasive computing environment.The system uses cluster-based local out-lier factor(CBLOF)detection method to detect data outliers,convolutional neural network attention-based long short-term mernory(CNN-ALSTM)model for intrusion detection and classification,and poor and rich optimization algorithm(PROA)to optimize CNN-ALSTM hyperparameters.Experimental results on benchmark datasets such as KDD CUP99,NSL-KDD,UN-SW-NB15 and CICIDS2017 show that the proposed intrusion detection system obtains an average of 98.77％precision and 98.36％recall on the binary classification task,and 96.87％precision and 92.36％recall on the multi-classification task,with the highest binary classification precision of 98.73％.The results are all better than other deep learning models.关键词
入侵检测系统/深度学习/异常检测/元搜索/智能环境Key words
intrusion detection system/deep learning/anomaly detection/meta-search/intelligent environment分类
信息技术与安全科学引用本文复制引用
曹俊捷,董会杰,方思敏..基于异常检测与优化深度学习的入侵检测系统[J].微型电脑应用,2025,41(2):9-12,4.基金项目
2022年度上海开放远程教育工程技术研究中心项目(KFKT2211) （KFKT2211）
