Application Research Of Data Mining Technology For Wearable Network

With the development of society,the usage of smart wearable devices is increasing gradually in these years.A wide variety of wearable devices in wearable networks produce a large amount of data that is mass,diverse,and valuable.It can inject new vitality into the development of the wearable network and serve the users better by using data mining related technology to analyze the relevant data in the wearable network and extract valuable information from these data.This paper use data mining related technology to study and analyze the user's gait data and the physiological sign during exercise in the wearable network,so as to further expand the related functions of wearable network.For the gait data of users in the wearable network,we firstly propose a complete user feature extraction scheme by using data mining related technology,and form a legal user identity comparison library.On the foundation of establishment of the legal user identity comparison library,we use the relevant anomaly detection model in data mining technology to detect the illegal users.On the basis of traditional classification algorithm,a joint voting classification model is designed to realize accurate identification of legitimate user identity tags.Finally,a new user identity authentication mechanism is established based on user gait data.In order to verify the reliability of the new user identity authentication mechanism proposed in this paper,we use smart phones to collect gait data of different users,so as to compare and evaluate related models.The simulation results show that the proposed scheme not only can accurately detect abnormal users who are not in the identity comparison library,but also can accurately identify the identity of legitimate users.In the study of physiological sign of users during exercise,we analyze the change rule of physiological sign and motion data of human beings during exercise,and use Bayesian combined model to realize multi-step prediction of physiological sign during human exercise.In order to eliminate the accumulated error generated in the multi-step prediction process,we propose a accumulated error correction mechanism based on the Naive Bayesian model to achieve timely correction of the accumulated error generated in the multi-step prediction process.Finally,based on the user's motion data,a set of accurate multi-step prediction mechanism for physiological sign was established.Similarly,we design related experiments and collect data of different exercisers during exercise to evaluate the performance of the prediction mechanism we proposed and compare it with other prediction mechanisms.The simulation results show that the proposed prediction mechanism has higher prediction accu-racy and lower prediction error than other prediction mechanisms.