To search, Click below search items. |
| All Published Papers Search Service |
|
|
|
Title |
An Effective Deep Autoencoder Approach for Online Smartphone-Based Human Activity Recognition |
|
Author |
Bandar Almaslukh, Jalal AlMuhtadi, and Abdelmonim Artoli |
| Citation |
Vol. 17 No. 4 pp. 160-165 |
|
Abstract |
Smartphones based human activity recognition (HAR) has a variety of applications such as healthcare, fitness tracking, etc. Nowadays, the signals generated by smartphone-embedded sensors such as accelerometer and gyroscope are used for HAR. However, achieving high recognition accuracy with low computation cost is required in smartphone based HAR. Therefore, we utilize one of the well-known deep learning approach named stacked autoencoder (SAE) to enhance the recognition accuracy and decrease recognition time. To evaluate the proposed method, we applied it on a public benchmark dataset and compared it against available methods known of highest recognition accuracy on the same dataset. We have found that the new method increase the overall classification accuracy from 96.4% to 97.5% and as well the average recognition time of each testing sample is decreased from 0.2724ms to 0.0375ms. |
|
Keywords |
Deep Learning, Stacked Autoencoder, Human Activity Recognition, Machine Learning. |
|
URL |
