摘要
针对大多数深度学习算法只使用单一模态进行分类会导致结果具有偏差性的问题,提出了一种基于双模态特征的混合神经网络。该方法能够使用两种不同的模态训练分类模型,提高分类模型的准确率。首先使用传输层流量数据包的有效载荷特征作为数据包级模态,数据包的长度序列特征作为流级模态;其次分成两个路径使用神经网络分析双模特征;再次将两条路径提取的高维特征进行融合;最后输出模型的分类结果。分别使用两个公开数据集对模型进行训练和测试,实验结果表明,多模态模型的分类精确率分别达到96.46%和93.01%,与当前4种比较优秀的单模态和多模态方法相比,均有明显提升。
To address the problem of biased results caused by most deep learning algorithms using a single modality for classification,this paper proposes a hybrid neural network based on bimodal features.The method can use two different modalities to train the classification model and improve the accuracy of the classification model.First,the payload features of the transport layer traffic data packets are used as the packet-level modals and the length sequence features of the data packets are used as the flow-level modals.Then,a neural network is used on two paths to analyze the bimodal features,and the high-dimensional features extracted from the two paths are fused.Finally,the classification results of the model are output.Two public datasets are used to train and test the model.Experimental results indicate that the classification precision of the bimodal model reaches 96.46%and 93.01%,respectively,which is significantly improved compared with the current four excellent single-modal and multi-modal method.
作者
田鑫
丁要军
TIAN Xin;DING Yaojun(Gansu University of Political Science and Law,Lanzhou Gansu 730070,China)
出处
《通信技术》
2023年第11期1267-1274,共8页
Communications Technology
