Abstract: Human Activity Recognition (HAR) is considered a challenging task in sensor-based monitoring systems. In ambient intelligent environments, such as smart homes, collecting data from multiple sensors is useful for recognizing Activities of Daily Living (ADLs), which can then be used to help provide assistance to inhabitants. ADLs are composed of complex time-series data that has high dimensionality, is large in size, and is updated continuously. Thus, developing methods for analysing these time-series data to extract meaningful features and specific characteristic would help solve the problem of activity recognition. Based on the noticeable success of deep learning in the time-series classification field, we developed a model for classifying ADLs in an ambient environment using deep neural networks. Our model, a Deep One-Dimensional Convolutional Neural Network (Deep 1D-CNN), contains several one-dimensional convolution layers coupled with a max-pooling technique to discover and extract the suitable internal structure to generate the deep features of the input time-series automatically. Such a model can be used as a unified framework for both feature extraction and classification. It performs well on high-dimensional time-series data; it does not require any expert knowledge in feature extraction, and it is able to find relevant and discriminative features for activity recognition. In order to evaluate the performance of our model, we tested it on the new real-life dataset, ContextAct@A4H, and the results showed that our model achieved a high F1 score (0.90). We also compared our results with baseline models for time series classification with deep neural networks. The comparison revealed that, our deep 1D-CNN model achieved the best overall performance in terms of precision, recall, and F1 score.


Keywords: Deep Learning, One-Dimensional Convolutional Neural Networks, Time-series Classification, Activities of Daily Living (ADLs), smart home

PDF | DOI: 10.17148/IJARCCE.2020.9301

Open chat
Chat with IJARCCE