About the project
Sensors and embedded systems generate high-frequency and high-volume time series. For example, smart watches generate a constant stream of motion data. Developing algorithms for machine learning from time series is an active research area.
Streaming sensor data give rise to problems in classification (for example, does this motion correspond to a specific action), anomaly detection (for example, has a person fallen over) and change point detection (for example, moving from a resting to a standing position).
Many of the algorithms to address the time series specific problems are computationally intensive and recovering all the sensor data to apply the learning algorithm is usually impractical: our experiments with smart watches have shown they rapidly heat up and become unwearable if data is extracted directly.
This PhD project will explore whether we can develop simplified on-device algorithms that can make the decision whether to send data to central control/cloud for testing with big model. The research question is then how we get the best trade-off between accuracy, sensitivity or specificity with lowest power footprint for classification and anomaly detection problems.
There are many potential case studies for this research project. Some examples of possible application are
- insect detector devices
- human activity recognition
- or device fault detection
This project could involve collaborations and placements with researchers in California, Greece, Kenya, France, Ireland, or Australia.
If you wish to discuss any details of the project informally, please contact Tony Bagnall, VLC Research Group, Email: mindscdt@soton.ac.uk
UKRI MINDS CDT
This project is funded through the UKRI MINDS Centre for Doctoral Training (www.mindscdt.ai). This is one of 16 PhD training centres in the UK with a unique focus on advancing AI techniques in the context of real-world engineered systems with a remit that spans novel hardware for AI, AI and machine learning, pervasive systems and IoT, and human-AI collaboration. We provide enhanced cross-disciplinary training in electronics and AI, entrepreneurship, responsible research and innovation, communication strategies, outreach and impact development as part of an integrated 4-year iPhD programme.