Pattern Recognition Techniques for the Identification of Activities of Daily Living Using a Mobile Device Accelerometer

• DOI: 10.3390/electronics9030509
• Publication Date: 2020-03-20T11:29:07Z
• Authors: Ivan Miguel Pires; Gonçalo Marques; Nuno M. Garcia; Francisco Flórez-...; Maria Canavarro T...; Eftim Zdravevski; Susanna Spinsante; Miguel Coimbra
• Keywords: Overfitting; Identification; Activity Recognition

The application of pattern recognition techniques to data collected from accelerometers available in off-the-shelf devices, such as smartphones, allows for the automatic activities daily living (ADLs). This can be used later create systems that monitor behaviors their users. main contribution this paper is use artificial neural networks (ANN) ADLs with acquired sensors mobile devices. Firstly, before ANN training, device collection. After devices are apply an previously trained ADLs’ identification on a less restrictive computational platform. motivation verify whether overfitting problem solved using only accelerometer data, which also requires resources and reduces energy expenditure when compared multiple sensors. presents method based defined set ADLs. It provides comparative study different implementations choose most appropriate identification. results show accuracy 85.89% deep (DNN).