Meta-learning, also known as “learning to learn”, is a machine learning technique that aims to design models that can learn from new environments rapidly with a very limited number of samples. In recent years, there has been rapid development in the meta-learning field . The goal of meta-learning is to enable machines building generalisable models that can be adaptive to different unseen datasets with few extra trainings. Meanwhile, rapid development of Internet of Things (IoT) enables modern smart distributed systems such as smart home network, smart car network, etc. Smart distributed systems are expected to significantly reduce the amount of computation in central ‘masters’ (e.g. servers in the remote cloud) and the communication cost between remote devices by enabling the majority ‘slave’ agents to accomplish the basic computational tasks. However, practical scenarios are often complicated, and the actual tasks for agents in different environments could be very different. Questions such as ‘how to efficiently enable a specific agent to learn from newly observed data’, and ‘how to update new learnt knowledge to all other agents in an adaptive way’ were raised in recent years.
This PhD project aims to resolve the above challenges by developing reliable and efficient statistical meta-learning techniques for applications in the context of smart car network or smart city, etc. Outcomes of this project are expected to improve our understanding of intelligibility and reliability in heterogeneous intelligent systems. This project also aims to explore feasible routes leading to next generation of smart services by using meta-learning enabled generalised distributed AI systems. This project might involve reliability examination of the newly learnt knowledge (from agents), risk and quality assessment for new knowledge updating, and development of novel and efficient meta-learning algorithms in the distributed environment.
The project will contribute to the exploration of safe and trustworthy AI research, and to improve the reliability and efficiency of smart cities, autonomous vehicles, automated industrial assembly, etc.
This project is associated with the UKRI CDT in Accountable, Responsible and Transparent AI (ART-AI), which is looking for its second cohort of at least 10 students to start in September 2020. Further details can be found at: http://www.bath.ac.uk/centres-for-doctoral-training/ukri-centre-for-doctoral-training-in-accountable-responsible-and-transparent-ai/
Applicants should hold, or expect to receive, a First or Upper Second Class Honours degree in Statistics, Computer Science, Mathematics, Information Engineering, or a closely related discipline. A master’s level qualification would be an advantage. A good mathematical or statistical academic profile, or good coding skill is a plus. Prior knowledge in machine learning is desirable, but not required.
Informal enquiries about the project should be directed to Dr Xi Chen: [email protected]
Enquiries about the application process should be sent to [email protected]
Formal applications should be made via the University of Bath’s online application form: https://samis.bath.ac.uk/urd/sits.urd/run/siw_ipp_lgn.login?process=siw_ipp_app&code1=RDUCM-FP02&code2=0002
Start date: 28 September 2020.
ART-AI CDT studentships are available on a competition basis for UK and EU students for up to 4 years. Funding will cover UK/EU tuition fees as well as providing maintenance at the UKRI doctoral stipend rate (£15,009 per annum in 2019/20, increased annually in line with the GDP deflator) and a training support fee of £1,000 per annum.
We also welcome all-year-round applications from self-funded candidates and candidates who can source their own funding.
 Finn, C., Abbeel, P., & Levine, S. (2017). Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks. In International Conference on Machine Learning (ICML), pp. 1126-1135.
 Edwards, H., & Storkey, A. (2016). Towards a neural statistician. arXiv preprint arXiv:1606.02185.