Activity recognition using smart phone acceleroeter with naive Bayes classifier for emergency cases /
Activity is one of the main components of context-aware study besides time, location and identity. Accurate recognition of user activity allows a device or application to deliver more accurate response to the user based on what it is designed for. Accelerometer has become one of the commonly used se...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
Gombak, Selangor :
Kulliyyah of Information and Communication Technology, International Islamic University Malaysia,
2016
|
| Subjects: | |
| Online Access: | Click here to view 1st 24 pages of the thesis. Members can view fulltext at the specified PCs in the library. |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Activity is one of the main components of context-aware study besides time, location and identity. Accurate recognition of user activity allows a device or application to deliver more accurate response to the user based on what it is designed for. Accelerometer has become one of the commonly used sensors for recognising user activity, especially with the recent availability of the sensor in smart phones. One key issue that arises together with accelerometers' popularity is mainly the performance in recognising the activity, which is partially influenced by the classification algorithm used. Thus, this research is trying to look into the matter by evaluating the performance of the classifiers for activity recognition using smart phones. Instead of focusing solely on the classification accuracy, this research extends the offline performance evaluation to include other evaluation measures like precision, recall, F-measure (the weighted harmonic mean of the precision and recall), and Receiver Operating Characteristic (ROC) area so that a non-bias evaluation is acquired, and issue like accuracy paradox can be avoided. Other aspects that might influence the performance like the size of training data, sensor location and gender are also explored. Sample applications are further developed in order to ensure the performance of the shortlisted classifier is consistent when implemented on smart phones for real-time/ online activity recognition, given the limited resources and computing capability. Finally, the finding from the study is then implemented in an activity-aware emergency alert application to demonstrate how the activity context retrieved can contribute to creating a more context-aware and ubiquitous environment. |
|---|---|
| Physical Description: | xiv, 133 leaves : ill. ; 30cm. |
| Bibliography: | Includes bibliographical references (leaves 113-116). |