Agent技术下分布式网络混合人侵信息智能检测

打开文本图片集

关键词：入侵信息检测；Agent技术；分布式网络；主成分分析；特征向量；多头注意力机制中图分类号：TN915.08-34；TP309 文献标识码：A 文章编号：1004-373X（2025）20-0176-05

Intelligent detection ofmixed intrusion informationin distributed networks under Agent technology

LI Xiaosong （School ofComputer Science，China WestNormal University，Nanchong637oo2，China）

Abstract：Thedistributednetworks havedecentralizedcharacteristics，withnosinglecentral nodecontroling theentire network.Duetothecharacteristicofdatabeingstoredinmultiplenodesinadecentralizedmanner，hichincreasesthedificulty ofdetectingmixedintrusioninformation.Therefore，amethodofdistributednetworkmixedintrusioninformationintellgent detectionbasedonAgenttechnologyisproposed.Themulti-Agenttechnologyisusedtodistributedetectiontaskstovarious units，aggregatingdatainformationfromthedistributednetwork，andextracting featurevectorsbymeansofprincipalcomponent analysis.ThefeaturevectorisusedasinputsofintrusiondetectionmodelbasedonCNN-BiLSTMtocapturetheirtemporal featuresbymeansoftheBiLSTMlayer，andintegratetheCNNandBiLSTMlayersbymeansof multi headatentionmechanism tooutputintrusiondetectionresults，soastorealizeinteligentdetectionofmixedintrusioninformationindistributednetworks. Theexperimentalresultsshowthattheproposedmethodcanachieveintrusioninformationdetectionbyselectingkeyfeaturesof distributed network information.

Keywords：intrusioninformationdetection；Agenttechnology；distributednetwork；principalcomponentanalysis; eigenvector;multi-head attention mechanism

0 引言

网络的开放性和共享性在促进信息交流的同时，也带来了严峻的安全挑战。（剩余5774字）