改进强化学习的AI远程终端用户身份识别仿真

AI Remote End User Identification Simulation Based on Improved Reinforcement Learning

受限于AI技术以及远程智能终端网络条件的复杂性,终端用户的网页浏览行为跟踪过程易产生冗余数据,用户身份识别难度较大。为此,提出基于强化学习的AI远程终端用户身份识别方法。从解锁行为、操作行为、通信行为等方面判断远程终端用户行为规律,在客户端中通过用户ID、访问页面地址、页面标题等属性定义用户终端浏览行为。将浏览信息传输至中心服务器并录入终端数据库内,采集完整终端用户数据。通过小波阈值方法消除冗余信息,根据强化学习的奖励持续调节方法,提取AI远程终端用户行为数据集,计算用户身份特征与行为特征间的耦合关系,得到身份识别结果。仿真结果表明,所提方法能够快速准确地识别目标用户身份,保障了用户数据安全,为其提供更可靠的AI远程操作环境。

Due to the complexity of AI technology and the network conditions of remote intelligent terminals,it is easy to generate redundant data in the process of web browsing behavior tracking.Therefore,a method of identifying AI remote terminal users is proposed based on reinforcement learning.Firstly,we judged the behavior law of remote terminal users from the aspects of unlocking behavior,operation behavior and communication behavior,and then defined the browsing behavior of user terminals through user ID,web page address,web page title and other attributes in the client.After that,we transmitted the browsing information to the central server and put it into the terminal database.Moreover,we collected complete end-user data.Furthermore,we used the wavelet threshold method to eliminate redundant information.According to the incentive adjustment method based on reinforcement learning,we extracted the behavior data set of AI remote terminal users,and calculated the coupling relationship between user identity characteristics and behavior characteristics.Finally,we got the recognition result.Simulation results show that the proposed method can identify target users quickly and accurately and ensure the security of user data,thus providing a more reliable AI environment.