Scalable and Interpretable Time Series Machine Learning for Industrial Systems

Dr Matthew Middlehurst, Prof Savas Konur

Applications accepted all year round

Self-Funded PhD Students Only

About the Project

Modern manufacturing, engineering and automotive systems generate vast volumes of time series data at high speed. Many leading time series classification (TSC) and regression (TSR) methods [1,2] perform well on the relatively small datasets in available benchmarks [3] but struggle as data grows in the number of cases, channels and/or series length. This limits their applicability when models must run under strict memory, energy and latency constraints at the edge. Improving these algorithms will open the door to a range of applications with potential to increase industrial efficiency.

This project tackles that gap by developing TSC and TSR techniques that scale to big data and run efficiently on-chip/at the edge, making them practical for real deployments. The emphasis is on maintaining high predictive accuracy while reducing training and inference costs, enabling use in resource-constrained, time critical settings. Results will be validated on large benchmarks [4] and realistic data streams. An initial aim for the project may include improving the scalability of ensemble approaches for TSC/TSR e.g., HIVE-COTE [5].

To illustrate impact, successful applicants will explore representative use cases such as predictive maintenance and prognostics in vehicles and industrial systems, optimising manufacturing processes such as moulding, and monitoring smart materials/structures. There will be opportunity for local and international collaboration throughout the project. Where possible, the student will leverage the University of Bradford’s engineering experience and collection of industrial datasets to ground the work in real data. Successful applicants will engage with an international network of TSML researchers based around the open-source aeon toolkit to support reproducibility and uptake. Depending on applicant interest, there may be scope for additional/alternative applications such as digital health and human activity recognition using wearable devices.

How to apply

Formal applications can be submitted via the University of Bradford web site https://evision.bradford.ac.uk/urd/sits.urd/run/siw_ipp_lgn.login?process=siw_ipp_app_crs. Applicants should register an account, select 'Postgraduate Research' as the course type and use the keywords 'computer science'. Please include the project title on the Research Proposal section; applicants are not required to supply a research proposal for this project.

Informal enquiries are also welcome.

About the University of Bradford

Bradford is a research-active University supporting the highest-quality research. We excel in applying our research to benefit our stakeholders by working with employers and organisations world-wide across the private, public, voluntary and community sectors and actively encourage and support our postgraduate researchers to engage in research and business development activities.

Funding Notes

This is a self-funded PhD project; applicants will be expected to pay their own fees or have a suitable source of third-party funding. UK students may be able to apply for a Doctoral Loan from Student Finance for financial support.

References

[1] Middlehurst, M., Schäfer, P. and Bagnall, A., 2024. Bake off redux: a review and experimental evaluation of recent time series classification algorithms. Data Mining and Knowledge Discovery, 38(4), pp.1958-2031. [2] Guijo-Rubio, D., Middlehurst, M., Arcencio, G., Silva, D.F. and Bagnall, A., 2024. Unsupervised feature based algorithms for time series extrinsic regression. Data Mining and Knowledge Discovery, 38(4), pp.2141-2185. [3] Dau, H.A., Bagnall, A., Kamgar, K., Yeh, C.C.M., Zhu, Y., Gharghabi, S., Ratanamahatana, C.A. and Keogh, E., 2019. The UCR time series archive. IEEE/CAA Journal of Automatica Sinica, 6(6), pp.1293-1305. [4] Dempster, A., Foumani, N.M., Tan, C.W., Miller, L., Mishra, A., Salehi, M., Pelletier, C., Schmidt, D.F. and Webb, G.I., 2025. MONSTER: Monash Scalable Time Series Evaluation Repository. arXiv preprint arXiv:2502.15122. [5] Middlehurst, M., Large, J., Flynn, M., Lines, J., Bostrom, A. and Bagnall, A., 2021. HIVE-COTE 2.0: a new meta ensemble for time series classification. Machine Learning, 110(11), pp.3211-3243.