摘要
针对现有安全事件预测算法所存在的过分依赖数据包头信息、所需历史数据较多、预测值误差较大、易陷入局部最优、训练时间较长等缺点,本文提出了一种基于递归神经网络进行分析数据包及其有效负载而在攻击发生前对安全事件进行预测的算法。该算法首先从数据包中提取源IP地址、协议类型和有效负载作为递归神经网络模型的输入,之后采用训练集对模型进行训练,同时引入批量梯度下降更新模型参数,最后采用测试集评估模型预测的准确率。通过递归神经网络分析有效负载可以更准确判断攻击性,大幅度提升网络安全事件的预测精度。
Aiming at the shortcomings of the existing security event prediction algorithm, such as over - reliance on data packet header information, more historical data required, large error of prediction value, easy to fall into local optimum and training time is longer. This paper proposes a new method based on recurrent neural network to analyze packets and their payloads and to predict security e- vents before an attack occurs. Firstly, the source IP address, protocol type and payload are extracted from the data packet as the input of recursive neural network model. Then the training set is used to train the model. At the same time, the model parameters are upda- ted by the batch gradient descending. Finally, the test set is used to evaluate the prediction accuracy of the model. Through the recur- rent neural network analysis of the payload can be more accurate to determine the attack, greatly enhance the network security event prediction accuracy.
出处
《网络新媒体技术》
2017年第5期54-58,共5页
Network New Media Technology
基金
中国科学院先导专项(XDA06040602)
临港地区智能制造产业专项(ZN2016020103)
关键词
安全事件预测
递归神经网络
有效负载
Network Security Event Prediction, Recurrent Neural Network, payload
