For curbing the global prevalence of physical inactivity and the associated average 5.3 million deaths per year, the importance of the Physical Activity (PA) research has been demonstrated by Word Health Organisation (WHO) reports. UK estimates in 2017 suggest over a quarter of people aged 16 years and over were categorised as ‘physically inactive’. Research has shown that the advent of wearable devices like wristbands or smartphones enables tracking and managing personal daily PA effortlessly, and potentially improve their health outcomes. But these wearable technologies suffer from low accuracy and weak robustness of objectively qualifying PA in free-living environments due to shortage of cost-effective wearable sensors, unstandardised baseline data, feasible machine learning algorithms, etc. This project addresses two key areas: i) Lack of feasible machine learning techniques for effectively feature extraction from unlabelled and uncertain PA data with high accuracy. ii) Lack of useful data mining and fusion techniques for improving accuracy and robustness of qualification of PA behaviour in long time free-living environments.
The main aim is to investigate how to design and develop cost-effective wearable system comprising advanced machine learning algorithms to objective quantification of a wide variety of personal physical activity (PA) behaviour in free-living environment. This will be evaluated by user studies in scenarios related to improving quality of life. The specific objectives are:
1. To conduct a state-of-the-art review of existing wearable technologies for monitoring and identifying PA in both controlled and free environments.
2. To study novel deep learning techniques that support automatic extraction of intrinsic and reliable features from unsupervised wearable sensing samples for accurate PA recognition.
3. To explore new data fusion approaches using large-scale life-logging PA data for improved accuracy and robustness of qualification of PA behaviour.
4. To undertake a case study of quantifying PA behaviour in real free-living environment to test, validate and demonstrate the developed wearable PA baseline, feature extraction algorithm, data fusion model and entire wearable intelligence systems.
Within this project you will extend our toolbox of human activity recognition methods by designing new machine learning and AI techniques tailored to wearable sensor data. Your research will contribute to extending our existing mobbing sensing platform Active 10, with over 650K active users. You will characterise the ability of the resulting system at recognising a wide range of daily physical activities through user studies both at the University and with our industrial partner.
As part of this project, you will stay in the OAK (Organisations, Information and Knowledge research group, which has acquired funding from EPSRC, EU-FP7, EUH2020, Innovate UK. The focus of our lab is on large-scale data and information acquisition, management and exploration. We have developed several mobile sensing platforms and software frameworks for tracking and monitoring human activity and mobility. The members of the lab have an international outlook, with a mix of computer scientists, computer engineers, and electronic engineers.
We are seeking an enthusiastic individual to join the OAK research group at Sheffield University, with the following attributes:
• A minimum 2.1 undergraduate (Bsc) and/or postgraduate masters’ qualification (MSc) in a science and technology field: Computer Science, Engineering, Mathematics, with specialisation in mobile computing, pervasive healthcare, machine learning and AI
• Familiarity with machine learning and probabilistic models
• Relevant software knowledge and experience, for example Python and tensor frameworks (PyTorch or TensorFlow), Jave, C++, etc
• Excellent analytical and numerical skills
• A driven, professional and independent work attitude
• Ability to liaise with academic supervisors from a range of disciplines
• Excellent written and verbal communication skills
If English is not your first language, you must have an IELTS score of 6.5 overall, with no less than 6.0 in each component.
To apply for the studentship, applicants need to apply directly to the University of Sheffield using the online application system. Please name Po Yang as your proposed supervisor.
Complete an application for admission to the standard Computer Science PhD programme https://www.sheffield.ac.uk/postgraduate/phd/apply/applying
Applications should include a research proposal, CV, transcripts and two references.
The research proposal (up to 4 A4 pages, including references) should outline your reasons for applying for this scholarship and how you would approach the researching, including details of your skills and experience.