Open Access Open Access  Restricted Access Subscription Access

Comparative Analysis of Recurrent Neural Network Architectures and Hyperparameters for Human Activity Recognition Using Wearable Sensors"


Affiliations
1 Research Scholar, Programmer, I. K. Gujral Punjab Technical University, Kapurthala, India
2 Assistant Professor, I. K. Gujral Punjab Technical University, Kapurthala, India
 

Human activity recognition (HAR) is a significant area of research with numerous applications in healthcare, athletics performance monitoring, and elderly care. The potential for HAR with ubiquitous sensors to enhance human performance and quality of life has attracted significant research interest. Recurrent Neural Networks (RNNs) have emerged as a powerful tool for HAR, as they can model sequential data and capture temporal dependencies in time-series data. Using accelerometer data from wearable sensors, this study investigates the efficacy of various recurrent neural network (RNN) architectures and hyperparameters for HAR. Specifically, we compare the performance of three RNN architectures (Simple RNN, LSTM, and GRU) and investigate the impact of hidden units and sequence length on the accuracy of the models. We use the publicly available HARUS dataset, which consists of accelerometer data collected from 30 subjects performing six different activities. Our results show that the LSTM architecture outperforms the other two architectures, achieving an accuracy of 95.0% on the HARUS dataset. We also discover that increasing the number of hidden units generally improves accuracy, with 128 hidden units producing the greatest results. Increasing the sequence length also leads to higher accuracy, but increasing it beyond a certain point can lead to overfitting. In addition, a separate study found that their RNN model obtained an overall accuracy of 99.54 percent on the test set for recognizing various activities using accelerometer data from a wearable sensor. The model performed particularly well for walking and jogging activities, as well as standing and sitting activities, and performed reasonably well for more complex activities such as walking upstairs and downstairs. Our findings indicate that LSTM is a suitable architecture for HAR tasks and that the number of concealed units and sequence length are crucial hyperparameters to consider. Our findings contribute to the existing literature on HAR by revealing the optimal architecture and hyperparameters for the accurate recognition of human activities from accelerometer data collected by wearable sensors.

Keywords

Human Activity Recognition, HARUS Dataset, Wearable Sensors, Accelerometer Data, Recurrent Neural Networks, RNN Architectures, Simple RNN, LSTM, GRU, Hyperparameters, Hidden Units, Sequence Length, Overfitting, Temporal Dependencies, Time-Series Data, Accuracy.
User
Notifications
Font Size

  • Chen, C., Wu, Y., & Zhang, Y. (2021). A comparative study of convolutional neural networks and recurrent neural networks for human activity recognition with wearable sensors. IEEE Access, 9, 49496-49507.
  • Gao, L., Li, P., Li, Y., Wu, S., & Chen, L. (2021). A novel hybrid model based on LSTM and attention mechanism for human activity recognition. Information Sciences, 564, 60-72.
  • Xia, Z., Ni, X., Liu, C., & Sun, J. (2021). Deep neural networks for human activity recognition based on smartwatch sensor data. Sensors, 21(1), 234.
  • Zhu, W., Qiu, J., Zhou, F., & Wan, Y. (2020). Human activity recognition based on wearable sensors using a hybrid CNN-LSTM model. IEEE Access, 8, 157801-157810.
  • Zhu, Y., Wu, X., & Liu, X. (2020). A hybrid CNN-LSTM model for human activity recognition. IEEE Access, 8, 55817-55827.
  • Zhang, J., Yin, J., & Chen, K. (2019). An improved CNN for human activity recognition using mobile phone sensor data. Sensors, 19(7), 1508.
  • Zhang, X., Zhou, W., & Chen, S. (2019). Human activity recognition based on wearable sensor data using convolutional neural networks. Sensors, 19(9), 2067.
  • Khan, A., Zhang, Y., & Ho, K. C. (2018). Human activity recognition and classification using machine learning algorithms. Sensors, 18(4), 1153.
  • Khan, A. M., Lee, Y. K., & Lee, S. Y. (2018). Human activity recognition via smartphone using SVM, KNN, and Naïve Bayes classifiers. Journal of Convergence, 9(4), 39-46.
  • Ordóñez, F. J., &Roggen, D. (2016). Deep convolutional and LSTM recurrent neural networks for multimodal wearable activity recognition. Sensors, 16(1), 115.
  • Padilla-López, J. R., Morales, J. A. G., & Fernández-Caballero, A. (2015). Feature selection for human activity recognition on smartphones using random forests. Sensors, 15(4), 7712-7730.
  • Padilla-López, J. R., Chaaraoui, A. A., &Flórez-Revuelta, F. (2015). Feature selection and reduction in accelerometer-based activity recognition. Sensors, 15(12), 30169-30189.

Abstract Views: 206

PDF Views: 0




  • Comparative Analysis of Recurrent Neural Network Architectures and Hyperparameters for Human Activity Recognition Using Wearable Sensors"

Abstract Views: 206  |  PDF Views: 0

Authors

Tarun Kanodia
Research Scholar, Programmer, I. K. Gujral Punjab Technical University, Kapurthala, India
Brijesh Bakariya
Assistant Professor, I. K. Gujral Punjab Technical University, Kapurthala, India

Abstract


Human activity recognition (HAR) is a significant area of research with numerous applications in healthcare, athletics performance monitoring, and elderly care. The potential for HAR with ubiquitous sensors to enhance human performance and quality of life has attracted significant research interest. Recurrent Neural Networks (RNNs) have emerged as a powerful tool for HAR, as they can model sequential data and capture temporal dependencies in time-series data. Using accelerometer data from wearable sensors, this study investigates the efficacy of various recurrent neural network (RNN) architectures and hyperparameters for HAR. Specifically, we compare the performance of three RNN architectures (Simple RNN, LSTM, and GRU) and investigate the impact of hidden units and sequence length on the accuracy of the models. We use the publicly available HARUS dataset, which consists of accelerometer data collected from 30 subjects performing six different activities. Our results show that the LSTM architecture outperforms the other two architectures, achieving an accuracy of 95.0% on the HARUS dataset. We also discover that increasing the number of hidden units generally improves accuracy, with 128 hidden units producing the greatest results. Increasing the sequence length also leads to higher accuracy, but increasing it beyond a certain point can lead to overfitting. In addition, a separate study found that their RNN model obtained an overall accuracy of 99.54 percent on the test set for recognizing various activities using accelerometer data from a wearable sensor. The model performed particularly well for walking and jogging activities, as well as standing and sitting activities, and performed reasonably well for more complex activities such as walking upstairs and downstairs. Our findings indicate that LSTM is a suitable architecture for HAR tasks and that the number of concealed units and sequence length are crucial hyperparameters to consider. Our findings contribute to the existing literature on HAR by revealing the optimal architecture and hyperparameters for the accurate recognition of human activities from accelerometer data collected by wearable sensors.

Keywords


Human Activity Recognition, HARUS Dataset, Wearable Sensors, Accelerometer Data, Recurrent Neural Networks, RNN Architectures, Simple RNN, LSTM, GRU, Hyperparameters, Hidden Units, Sequence Length, Overfitting, Temporal Dependencies, Time-Series Data, Accuracy.

References