Family Malicious Domain Name Detection Based on Intra-layer and Inter-layer Fusion Attention
A new method for detecting family malicious domain names based on intra-layer and inter-layer fusion attention is proposed to address the issues of low accuracy and high false positives in the detection of newly emerged or mutated malicious domain names using current methods.Firstly,it uses deep self-encoded networks to compress the domain name set layer by layer into spatial features,and utilizes self-attention mechanisms to enhance the expression ability of key characters in domain name strings.Secondly,it utilizes cross attention to establish associations between the input ends of a dual branch network,promoting the exchange of deep information between branches.Finally,it calculates the similarity comparison between the mapping features of domain name to be tested and the interaction feature set.The experimental results show that the accuracy of the designed method is 98.21%,which is of great practical significance for ensuring network security and preventing new domain name intrusion attacks.
malicious domain name detectionfusion attentionjudgment ruleintra-layer self-attentioninter-layer cross-attention
国家科技期刊平台
NSTL
万方数据
