Supported by the EIT Digital, City, University of London’s Artificial Intelligence Research Centre (CitAI) is offering a Doctoral Training Program (CeitIAI) in Industrial Artificial Intelligence. This program will enable the commercial deployment of AI technology in the UK, Europe and other regions by developing critical interdisciplinary capabilities in AI technology, particularly, though not exclusively in deep learning, explainable AI, and Artificial General Intelligence. The DTP is focused on the application of AI for solving real-life industrial problems in energy, transport, health and well-being, and finance as well as on the study of the legal, ethical and social impact of AI.
The EIT Digital School is a leading European digital innovation and entrepreneurial education organisation driving Europe’s digital transformation. It invests in strategic areas to accelerate the market uptake and scaling of research-based digital technologies (deep tech) focusing on Europe’s strategic, societal challenges: Digital Tech, Digital Cities, Digital Industry, Digital Wellbeing, and Digital Finance.
CeitIAI is funded through studentships sponsored by EIT-Digital and industrial partners. The studentships consist of a full fee waiver and a stipend of £18K per year, for four years. As part of their studies and training, the students spend time at City’s School of Mathematics, Computer Science and Engineering (SMCSE), EIT-Digital London Co-Location Centre and the premises of the company which co-funds the projects. In addition, over the 4 years of study, the PhD student follows the EIT Digital PhD, which is an enrichment programme to develop skills and competences in innovation and entrepreneurship in digital technologies. The label programme supports: 1 between 3 and 6 months abroad to enrich their research experience, for which a supplementary budget is available; 2 EIT-Digital European training in innovation, entrepreneurship and digital transformation leadership, which takes place at other EIT-Digital centres across Europe; 3 Business Development Experience, which is a series of activities planned together with the industrial supervisor to put in practice the knowledge acquired in the seminars.
At City, the PhD students are assigned academic supervisors, and have access to computing facilities, including the new cutting-edge AI Robin Milner Lab, which includes GPU computers and servers, and City’s services. In addition, they are monitored and supported by a Senior Tutor for Research, participate in SMCSE’s DTP, and gain teaching experience by supporting our UG and PG courses.
We are aiming to sponsor 10 studentships per year for the next 4 years, starting with 3 projects:
- Algorithms for predictive maintenance of vehicles in a connected environment: “Novel algorithms will be developed for predicting and specifying repair and maintenance requirements of vehicles in connected environments, where the vehicle and the repair workshop are in two-way data communication with cloud-based services. These may include algorithms for identifying what data to collect, identifying and predicting faults using collected data, advising vehicle operators of repair and maintenance requirements and optimising the repair and maintenance tasks to be performed”. Sponsored by Bosch AA-AS.
- Customisable Private Deep Learning: “Design and evaluation of end to end, secure, efficient privacy-preserving deep learning. Although current literature includes several proposals to run deep leaning with privacy-preservation, the performance is still far from satisfactory. This PhD project will focus on using current technology (homomorphic encryption/secure multi-party computation/differential privacy/federated learning) in a principled approach to design customised solutions for use cases, that improve the computational efficiency and performance of deep learning architectures under given privacy constraints”. Sponsored by Delta Capita.
- Efficient Privacy Preserving Scheme for Data Processing: “Design, implementation and evaluation of an efficient scheme which will preserve user privacy, reduce the complexity of processing noise and allow multi-user access for machine learning and other applications. Current literature focuses on fully homomorphic encryption schemes which allow single user access under a unique key and substantially high noise of multiplication depth. Processing results of encrypted data has shown huge leaps but still lack practicality as the data sets grow. The scalability of the current encryption schemes remains unpractical; current results range from 5 minutes for 4 l-bit elements to just under 2 hours for 16 l-bit elements. Even a small reduction in the overall multiplicative complexity can have a major impact on how the circuits scale. Reducing the multiplicative depth of homomorphically encryption algorithms specifically for deep learning architectures is essential”. Sponsored by Delta Capita.