Activity Recognition from Smartphones Using Hybrid Classifier PCA-SVM-HMM

Abstract

Physical activity recognition using embedded sensors has enabled many context-aware applications in different areas, such as healthcare. A various real-life ubiquitous computing applications use smart sensors embedded in smart phones to infer user's human activities. In this work, we proposed a new hybrid classification model to perform recognition of activities using Smartphone data. The proposed method combines SVM learning algorithm with HMM, to classify and identify activity. Principal Component Analysis (PCA) is used to reduce the features set in the dataset. Experiments performed in the real datasets show comparative results between this SVM-HMM, SVM, HMM and the baseline methods in terms of recognition performance, highlighting the advantages of the proposed method.