Ве молиме користете го овој идентификатор да го цитирате или поврзете овој запис:
http://hdl.handle.net/20.500.12188/14075
Наслов: | Homogeneous Data Normalization and Deep Learning: A Case Study in Human Activity Classification | Authors: | Pires, Ivan Miguel Hussain, Faisal M. Garcia, Nuno M. Lameski, Petre Zdravevski, Eftim |
Issue Date: | 10-ное-2020 | Publisher: | MDPI AG | Journal: | Future Internet | Abstract: | <jats:p>One class of applications for human activity recognition methods is found in mobile devices for monitoring older adults and people with special needs. Recently, many studies were performed to create intelligent methods for the recognition of human activities. However, the different mobile devices in the market acquire the data from sensors at different frequencies. This paper focuses on implementing four data normalization techniques, i.e., MaxAbsScaler, MinMaxScaler, RobustScaler, and Z-Score. Subsequently, we evaluate the impact of the normalization algorithms with deep neural networks (DNN) for the classification of the human activities. The impact of the data normalization was counterintuitive, resulting in a degradation of performance. Namely, when using the accelerometer data, the accuracy dropped from about 79% to only 53% for the best normalization approach. Similarly, for the gyroscope data, the accuracy without normalization was about 81.5%, whereas with the best normalization, it was only 60%. It can be concluded that data normalization techniques are not helpful in classification problems with homogeneous data.</jats:p> | URI: | http://hdl.handle.net/20.500.12188/14075 | DOI: | 10.3390/fi12110194 |
Appears in Collections: | Faculty of Computer Science and Engineering: Journal Articles |
Files in This Item:
File | Опис | Size | Format | |
---|---|---|---|---|
2020-11 futureinternet-12-00194-v2 data normalization.pdf | 800.61 kB | Adobe PDF | View/Open |
Page view(s)
73
checked on 19.4.2024
Download(s)
29
checked on 19.4.2024
Google ScholarTM
Проверете
Altmetric
Записите во DSpace се заштитени со авторски права, со сите права задржани, освен ако не е поинаку наведено.