FATIN AMIEZA BINTI MOHD PABLI
Fatin connects IT world with management to further organize computer system efficiently. At the USM, she learnt Distributed System and Security to further comprehend the vitality of data and privacy. It seems like she's the type to learn new things as she recently completed a research project on deep learning. It is a whole different than security field but she made it with the help of her supervisor. She does apply management in her activities and did online marketing during the pandemic. Fatin currently looking for a business analyst and uc?for any IT related position in the near future
Matrix No:
137064
Student Email:
Supervisor:
Dr. Mohd Halim Mohd Noor
Supervisor Email:
Deep Learning for Human Activity Recognition (Research Category)
RP002
Human activity recognition is an active field of research in computer vision. Traditionally, the method for human activity recognition is done by manually extract features which may hinder the generalisation model performance. Years before, many researchers have been focusing on deep learning method to recognise human activity with the use of wearable sensors. Convolutional Neural Network (CNN) is one of the deep learning methods that deliver promising results in human activity recognition. Although CNN has been shown to be effective in extracting relevant features, it is limited in capturing the temporal information in the time-series data. In this research, a hybrid of CNN and Recurrent Neural Network (RNN) with the variant of Long-Short Term Memory (LSTM) will be proposed to model the temporal relationship that have varies time span and signal distributions. The proposed model are then evaluated and compared with CNN using a public data set. The data set consist of twelve different types of activities that will be differentiated by two types of model which are CNN and the hybrid of CNN-LSTM,