基于循环神经网络的人体异常行为识别模型

Human abnormal behavior recognition model based on recurrent neural network

传统识别模型在进行人体异常行为识别时,无法准确地定位到识别目标的IP地址与目标源。针对此问题,设计了一种基于循环神经网络的人体异常行为识别模型。根据人体异常行为在循环神经网络中的传播方式,计算人体规律性异常行为、重复性异常行为在网络中占用的流量,并通过Lex.net技术提取网络规则,对比人体行为执行规则与循环神经网络规则,描述人体异常行为网络执行规则;同时,引进CNN标记方式,对异常信息进行标记,引入卷积神经网络标记异常信息,将标记结果按照高语义特征与低语义特征,以此实现对行为的有效识别。实验证明,本文设计的识别模型,可以在有限时间内输出所有人体异常行为,并能在完成对异常行为的识别后,追踪到行为对应的目标个体。

Traditional recognition model can not locate the IP address and source of the recognition target accurately when recognizing human abnormal behavior.In order to solve this problem,a recognition model based on cyclic neural network is designed.According to the propagation mode of human abnormal behavior in the circulating neural network,the traffic flow of human regular abnormal behavior and repetitive abnormal behavior in the network is calculated,and the network rules are extracted by Lex.net technology,and the execution rules of human behavior are compared with those of the circulating neural network to describe the execution rules of human abnormal behavior.At the same time,the CNN mark is introduced to mark the abnormal information,and the result is classified according to high semantic feature and low semantic feature.On this basis,it is proved by experiments that the model can output all the abnormal behaviors of human body in limited time,and can trace the target individual after recognizing the abnormal behaviors.