Most modern smartphones have various sensor types integrated and become a rich sensing platform for a wide range of applications including healthcare, sport, social media, environmental monitoring. In most applications, the sensory data depends heavily on the context of the smartphones when the data was being collected. For example, the Inertial Measurement Unit (IMU) data collected for human activity recognition are different when the smartphone in the user’s pocket or the user’s hands. Therefore, knowing the context will improve the sensing accuracy. Recently, it has been proved that transformer, an advanced deep learning technique for time-series data, can surpass the well-known Long-Short Term Memory (LSTM).
To this end, this project aims at exploiting the transformer to detect the context of smartphones.
The project will be divided into 3 main tasks:
- Study smartphone context detection with Convolutional Neural Networks (CNN) and Long-Short Term Memory (LSTM)
- Develop transformer for smartphone context detection
- Compare transformer with CNN and LSTM based fingerprinting localisation
30% Theory, 50% Implementation, 20%Writing
Le Viet Duc, email@example.com, room ZI 5013