Research And Application Of User Authentication Algorithm Based On Keystroke Feature

With the rapid development of information technology,people have raised higher demands on the quality of life,and the personal information security has become the focus of attention.The flood of data has enriched and facilitated people's lives,but it has also brought great challenges to the field of information security.Traditional authentication methods based on commands and passwords have been unable to meet the current security needs.In recent years,biometric identification technology as an emerging authentication method for computer access control has received more and more attention.In this dissertation,we design an identification model based on keystroke feature for the authentication requirement of computer system access control.Fitting a large number of keystroke feature data using Chaos particle swarm optimization algorithm to optimize the initial parameters of the BP neural network,by classifying and identifying different user keystroke characteristics to achieve the purpose of user authentication.1.Construct a keystroke feature identification model based on BP neural network.Analyze the behavior data characteristics generated by the user based on the static text keystroke operation,and extract the interval time and duration of keystroke generation as the characteristic parameters,and integrate the experiment data.Construct the neural network learning model,and classify the user's keystroke data by the data training model to realize the user identification.2.Using chaotic particle swarm optimization algorithm to optimize the initial parameters of the BP neural network,improve the identification model.Optimize the initial parameters of the network,avoid the learning process into local extremum,and effectively improve the learning efficiency,so that the identification model of BP neural network is applied to keystroke characteristics data is more reasonable,accurate and efficient.3.In the constructed identity model,the function and performance of the model are tested from many angles.The effectiveness and feasibility of the model are validated by comparing the learning effect and the recognition accuracy before and after the model optimization.The results of experiments show that the keystroke feature identification modelconstructed by optimized BP neural network can give the keystroke characteristics a accuracy classify,and realize the identification of the user.Which tests that the research method adopted in this dissertation is correct,and the designed network recognition model is effective.