A deep machine learning method for concurrent and interleaved human activity recognition

Abstract

Human activity recognition has become an important research topic within the field ofpervasive computing, ambient assistive living (AAL), robotics, health-care monitoring, and manymore. Techniques for recognizing simple and single activities are typical for now, but recognizingcomplex activities such as concurrent and interleaving activity is still a major challenging issue. Inthis paper, we propose a two-phase hybrid deep machine learning approach using bi-directionalLong-Short Term Memory (BiLSTM) and Skip-Chain Conditional random field (SCCRF) torecognize the complex activity. BiLSTM is a sequential generative deep learning inherited fromRecurrent Neural Network (RNN). SCCRFs is a distinctive feature of conditional random field(CRF) that can represent long term dependencies. In the first phase of the proposed approach, werecognized the concurrent activities using the BiLSTM technique, and in the second phase, SCCRFidentifies the interleaved activity. Accuracy of the proposed framework against the counterpartstate-of-art methods using the publicly available datasets in a smart home environment is analyzed.Our experiment’s result surpasses the previously proposed approaches with an average accuracyof more than 93%.