Abnormal traffic detection method based on multi-scale attention feature enhancement
To address feature redundancy and temporal dependencies in traffic data sequences that slow down model training and degrade performance of existing network abnormal traffic detection methods,an abnormal traffic detection method based on multi-scale attention feature enhancement was proposed.Firstly,an optimal feature set was selected from traffic data using a feature selection algorithm based on dynamic grouping.Secondly,Dense-CNN and a multi-scale attention feature extraction network were employed to extract local and global features of the traffic data.Finally,a fea-ture enhancement network was used to increase the distinctiveness and expressiveness of local and global features,which were then fused using a weighted fusion approach to achieve abnormal traffic detection.Experimental results on the CIC-IDS2017 and CSE-CIC-IDS2018 datasets show that the proposed method improves F1 score by 0.17%to 2.75%and 0.43%to 8.99%,respectively,which has good detection performance.
万方数据
维普
