The Human Activity Recognition using the Signal produced by the Sensors have a number of applications in field of the fitness and health. The Human activities are recorded with the help of the various types of the sensor embedded in a wearable device or in a smartphone. There are many research works have been done for the Human Activity Recognition using the machine-learning as well as deep-learning models, but there is requirement to find out that which model is more efficient for a specific dataset, for which the comparative study of the model comes in mind. In this research paper the comparative study of three most efficient Deep Learning models LSTM-RNN, GRU-RNN and CNN has been performed on the most famous dataset ‘Human Activity Recognition Using Smartphones Data Set’ present at UCI machine-learning repository. ‘LSTM-RNN’ is abbreviated for ‘Long Short-Term Memory-Recurrent Neural Network’ is an updated version of the recurrent neural network based on the concept of back-propagation, is capable of remembering the dependencies for comparatively longer time-span. ‘GRU-RNN’ is abbreviated for ‘Gated Recurrent Units-Recurrent Neural Network’ is also an updated version of the recurrent neural network based on the concept of back-propagation, with fewer parameters than LSTM-RNN. ‘CNN’ is abbreviated for ‘Convolutional Neural Network’ is a feed forward Neural network using Convolutional layers for feature-extraction and fully-connected layer for classification.

Human Activity Recognition (HAR), LSTM-RNN, GRU-RNN, CNN

[1] S. S. Anju, K. V. Kavitha, “Performance Evolution of Varoius Machine Learning Technique for Human Activity Recognition using Smartphone”, Vol.7, Issue.8, pp.316-319, 2019.
[2] S. R. Ramammurthy, N. Roy, “Recent trends in machine learning for human activity recognition- A Survey”, Wiley Interdisciplinary Reviews: Data Minining and Knowledge Discovery, 2018, doi: 10. 1002/widm. 1254.
[3] S. Wan, Q. Lianyong, X. Xiolong, T. Chao, G. Zonhua, “Deep Learning Models for Real-time Human Recognition with Smartphones”, Mobile Networks and applications, 2019, doi: 10. 1007/S11036. 019- 01445- x.
[4] G.A. Kani, P. Geetha, A. Gomathi, “Human Activity recognition using deep learning with Ggradient Fused Handcrafted Features and Categorization based on Machine Learning Technique”, International Journal of Computer Sciences and Engineering, Vol.6, Issue.7, pp.1-7, 2018.
[5] N. Geeta, E.S. Samundeeswari, “A Review on Human Activity Recognition System”, International Journal of Computer Sciences and Engineering, Vol.6, Issue.12, pp.825-829, 2018.
[6] M. Badshah, “Sensor-based Human Activity Recognition using Smartphones”, Sensors, 2019, doi: 10.31979/etd.8fjc-drpn.
[7] R. Dey, F.M. Salem, “Gate-Variants of Gated Recurrent Unit (GRU) Neural Networks”, journal of IEEE Acess,2017, doi: 10.1109/MWSAS.2017.8053243.
[8] W. Jiang, Z. Yin, and., “ Human activity recognition using wearable sensors by deep convolutional neural networks.”, In Proceedings of the 23rd ACM international conference on Multimedia (pp. 1307-1310), 2015.
[9] H. Cho, S.M. Yoon, “Divide and Conquer based 1D CNN Human Activity Recognition using test data Sharpening”, Sensors, 2018, 1055, doi:10.3390/s 18041005.