A New Framework Using PCA, LDA and KNN-SVM to Activity Recognition Based SmartPhone’s Sensors
Ihssene Menhour, Mrhamed Bilal Abidine, Belkacem Fergani · 2018
In this paper, we proposed a new model to perform automatic recognition of activities using Smartphones data from a gyroscope and accelerometer sensors. We target assisted living applications such as activity monitoring for the disabled and the elderly persons. The proposed method combine the Principal Component Analysis (PCA) or Linear Discriminant Analysis (LDA) for dimension reduction and KNN-SVM using K-Nearest Neighbors (KNN) with Support Vector Machines (SVM) allowing to better discrimination between the classes of activities. Several experiments performed with real datasets shows a significant improvement of our proposed approach in terms of recognition performance.