Application Of Data Mining Technology In Human Motion Recognition Research

With the continuous development of Internet technology,personal health information management also gradually became popular.Personal health information management as a kind of brand-new network information management mode,you can follow the needs of users,and more by the patient to manage their own health is linked to health information and medical data.Rise in the field of pattern recognition in recent years,many new research direction,human activity state recognition based on inertial sensor is one of the faster growing branch,to the maximum extent to meet the demand of intelligent human-computer interaction and real-time health monitoring,one of the most concern is the human movement state recognition based on the acceleration sensor,through the medical and health care,sports data monitoring and evaluation of energy consumption in many scenes of good performance more and more get the favour of people.At the same time,compared with based on target tracking and event recognition,computer vision based on the acceleration sensor way and without being limited by the scene,less energy consumption,low cost and easy promotion,and many other advantages.Based on the acceleration sensor mobile health data collected by the human body,applying data mining technology to the research work of human movement state recognition,using the time-frequency characteristics of acceleration data,synthetically and keep standing to the human body each stride frequency decomposition of movement posture to carry on the accurate judgment;And in the limited training focus adopted integrated learning algorithm based on attribute combination,to solve the problem of unbalanced data a semi-supervised learning,build the random forest model to ride a bicycle to the human body state is used to identify the machine,integrated learning algorithm is analyzed according to the experimental results have advantages and disadvantages.