Research On Falling Detection And Daily Activities Recognition Based On Machine Learning Algorithm

With the growing phenomenon of an aging population, an increasing number of older people are living alone for domestic and social reasons. Based on this fact, falling accidents become one of the most important factors in threatening the lives of the elderly. Therefore, it is necessary to set up an application to detect the daily activities of the elderly. It is the problem that the elderly have to be faced with and the society has to deal with. Using sensors seems to be capable of handling these problems, however, falling detection is difficult to recognize because the “falling” motion is an instantaneous motion and easy to confuse with others. This thesis comprehensively takes the falling detection and daily activities recognition into consideration and makes research on them.In this thesis, three data mining methods were employed on wearable sensors’ value; first which contains the continuous data set concerning eleven activities of daily living, and then an analysis of the different results was performed, such as walking, falling, sitting down, lying, etc. Not only could the fall be detected, but other activities could also be classified. In detail, three methods including Back Propagation Neural Network, Support Vector Machine and Hidden Markov Model are applied separately to train the data set. According to the results using three existing machine learning algorithms, the analysis is preformed on accuracy, running speed and time complexity, especially on the reason of low accuracy.What highlights the project is that a new idea is put forward, the aim of which is to design a methodology of accurate classification in the time-series data set. The proposed approach, which includes obtaining of classifier parts and the application parts, allows the generalization of classification. The preliminary results indicate that the new method achieves the high accuracy of classification, whose accuracy is 90%, and significantly performs better than other data mining methods in this experiment, reached by 40%. Furthermore, this thesis integrates the above method, which contains not only a set of training and application aspects, but also the algorithm(HCA) on classifiers’ application.