Machine Learning for optimisation of Industrial Processes (industrial CASE award with Siemens)

The aim of this PhD is to develop a model for similarity that facilitates the borrowing of pertinent information from one context to neighbouring contexts where information is sparse. This is related to the problem of transfer learning (a specific variant of machine learning) in computer science and can also be considered as a regression problem in a Bayesian statistics context.
The specific application that motivates the research is the optimisation of industrial processes (eg making paint or babyfood). Increasing productivity of such processes has an immediate impact on the profitability of the process, making it extremely important to identify even modest increases in productivity. The processes are typically composed of a number of components where the composition of components and the configuration of each component are specific to the process. However, multiple processes often use the same components albeit in slightly different configurations. While it is well understood how to optimise a process using historic data from that process, using data from other similar processes is less well understood. This is particularly important in the context of new processes, for which historic data is sparse (and can even be non-existent), but is also very relevant in contexts where data exists for a large number of processes and the potential for each process to benefit from experience of the other processes is therefore significant.
The PhD will therefore comprise three main strands of research on: constructing mathematical models for industrial processes in terms of their commonality of components and their configurations; articulating the similarity of components and thereby processes; optimising a process given data from both that process and other similar processes.
The PhD will also involve active engagement with Siemens’ Process Industries and Drives division, which has been heavily involved in formulating the scope of the PhD and which are co-funding the project. Siemens is a global innovation powerhouse in electrical engineering and electronics with over 350,000 employees worldwide and 14,000 in the UK. Siemens Process Industries and Drives has an established track record developing and installing equipment to continuously improve the reliability, safety, and efficiency of products, processes and plants. Around the world, Siemens provide future-proof automation, drive technology, industrial software, and services based on best-in-class technology. With manufacturing systems producing more data than ever before, Siemens have identified a potential step change in the productivity of processes that could result from the work of this PhD.
The PhD includes components that pull on Computer Science, Statistics, and Engineering and is at the intersection of these three academic disciplines. The successful applicant will have experience in one of these domains and will gain experience in the others. It is also anticipated that the successful applicant will gain experience of Big Data technologies (eg Hadoop and Spark) and that the project will enable the student to enhance valuable programming skills (eg in MATLAB and Java).
Prof Simon Maskell (see: www.simonmaskell.com) and Prof Karl Tuyls (https://www.liverpool.ac.uk/computer-science/staff/karl-tuyls/) will supervise the project at the University of Liverpool. Simon has a growing vibrant team that currently includes PhD students and post-docs with backgrounds in statistics, computer science, engineering, particle physics and psychology working on applications that span, for example, aerospace, cyber security, insurance, healthcare and robotics. Karl has a team of researchers helping him pursue his research interests in topics that include: robotics, swarm intelligence, multi-agent systems, reinforcement learning, and evolutionary game theory. The primary point of contact at Siemens will be Mr Paul Hingley. Engagement with and contribution to the state-of-the-art are anticipated and the student will gain international exposure by presenting at top-ranked conferences.
As this studentship is within the School of Electrical Engineering, Electronics and Computer Science, applications can be submitted for either Electrical Engineering and Electronics’ or the Computer Science Department, please clearly state the project title and supervisor. Please apply via The University of Liverpool website:
https://www.liverpool.ac.uk/study/postgraduate/applying/online/

The PhD will be funded for 4 years by an industrial CASE award. This includes all Tuition fees and includes a top-up of £4,500 per year over and above the stipend associated with a standard EPSRC-funded PhD (normally approx. £14,000 maintenance per year). To be eligible, applicants must be Home/EU students due to funding.
The closing date for applications is 30 November 2016. The PhD start-date is to be agreed with the successful applicant, but could be immediate.