Comparing HMM, LDA, SVM and smote-SVM algorithms in classifying human activities

Abstract

Accurately recognizing the rare activities from sensor network based smart homes for monitoring the elderly person is a challenging task. Typically a probabilistic models such as the Hidden Markov Model (HMM) and Linear Discriminant Analysis (LDA) are used to classify the activities. In this work, we demonstrate that discriminative model named Support Vector Machines (SVM) based on the Synthetic Minority Over-sampling Technique (Smote) outperforms HMM, LDA and standard SVM and it can lead to a significant increase in recognition performance. Our experiments carried out on multiple real world activity recognition datasets, consisting of several weeks of data.