الفهرس 
INTRODUCTIONOnly 14 pages are availabe for public view
 |  | from | 202 |  |  |
| | | from | 202 | | |
Abstract
Context awareness with services discovery can help to provide better Human Activity Recognition(HAR). This thesis has presented a proposed framework based on context awareness for services discovery. The main aim is to recognize human activities using Back Propagation Neural Network (BPNN) and Deep Neural Network (DNN). They applied on small, medium, and large datasets to evaluate the performance of BPNN and DNN.
First, the BPNN on small dataset had achieved high accuracy 99.99% in training and 99.72% in the test using trainbr. Moreover, the results on medium dataset showed that trainbr had achieved the high accuracy 99.78% in training and 98.56% in test.
Second, DNN algorithm has implemented on small, medium, and full three datasets using trainscg. The results showed that DNN had achieved high accuracy 99.9% in training and 97.74% in test for HAR than medium dataset which achieved 99.98% in training and 93.19% in the test. Also, the small dataset has achieved 100% in training and 90.56% in the test and this result less than the full dataset. Finally, the results were compared with the previous work on the same used dataset and find that current results improved the activities recognizing accuracy in both algorithms: BPNN and DNN.