Some news items below provide further information on recent work undertaken by the group
Sanjay Rakshit, the Head of R&D Transwap delivered an online guest lecture and provided industry insights on how to evaluate ML models in the context of financial applications
Machine learning models have been widely applied to fintech applications. Sanjay Rakshit, the Head of R&D Transwap, a fintech firm headquartered in Singapore and recently expanded into the UK, has delivered an online guest lecture and provided industry insights about how to evaluate ML models in the context of financial applications to MSc students in Advanced Technology for Financial Computing. Sanjay has over 20 years industry working experience and shared his experience on practical application of evaluation of Machine Learning models for fraud detection. The talk has been well received by students and inspired students' learning experience in applying their data modelling skills into real-world fintech problems. Dr. Tiejun Ma has also highlighted the excellent opportunity for Informatics students to meet and discuss with industry experts like Sanjay to enrich students' learning, which is part of MSc in Advanced Technology for Financial Computing training.
Bayes Welcomes TranSwap
Dr. Tiejun Ma, the Business Informatics Theme leader within Bayes and University of Edinburgh, highlighted that we are glad to facilitate fintech and AI related collaboration with Trahswap's global R&D division. The School of Informatics and University of Edinburgh has world-class reputation with academic of excellence in Fintech and AI research.
The Bayes Centre is delighted that TranSwap have chosen Edinburgh as part of their global R&D expansion. As the University of Edinburgh's representative innovation hub for Data Science & AI, we look forward to supporting the development of next-generation cutting-edge financial services solutions delivered by TranSwap
Abnormally behaviour detection in digital financial services using supervised forecasting methods
Digital technology revolutionises the payment and currency services. Particularly digital currencies with pseudonymity, provide opportunities for illegal financial activities. To regulate the use of digital transaction services and avoid crimes, it is vital to identify efficient ways to detect anomalous users and grasp their underlying pattern of actions. Related research of risk modelling in digital currency users mainly relied on manual effort and rule-based methods. It is the lack of further optimisation and detailed analysis of underlying patterns of data that made those methods only able to detect outliers. This exciting new project is led by Dr. Tiejun Ma (Business Informatics) with an industry partner--INNOWISE Ltd and Mr Chang Luo, a talented graduate from Beijing University and now a PhD student within AIAI and Informatics. The project also involves collaboration with partners from University of Oxford and the Turing institute to explore the latest graph learning modelling techniques applied to abnormal events forecasting. The research team will explore new machine learning models combined with online data, which can identify potential risky transaction activities which improve industry financial risk management activities. The outcomes of this project will likely improve industry partners’ understanding about how to manage risks associated with clients’ activities and how to improve the prediction of risky events and alert the risk managers to further investigate these events. Due to the vast amount of the events on daily bases, this piece of research work would greatly reduce the effort and associated costs related to investigating risky transactions, improve regulation compliance, and reduce related risks and costs of running businesses.
Data Driven Innovation Fellowship: High performance data-driven trusted financial systems for next generation smart mobile users
5G technologies enable large-scale and distributed smart mobile devices to join in the network, and enables fast growth of new smart mobile services such as decentralised financial applications. Dr.Xiao Chen, a post-doc research fellow in AIAI, has been awarded a three years prestigious Data-Driven Innovation Fellowship (EU Marie Skłodowska-Curie) by DDI to address such challenges. The project is supervised by Dr. Tiejun Ma (Reader in Business Informatics) and together with Polydigi Ltd, a fintech firm with award-winning new mobile technologies with Smart City Award 2018 and a RBS award fintech startup. This project will particularly explore the synergies of academic expertise within AIAI and data innovations with commercial potential. The project will explore smart mobile technologies and client-facing trusted systems with large-scale mobile users. The research team will explore integrated data-driven design to optimise the performance of the next generation mobile services for banks, trading and insurance firms with trustworthy distributed transaction technologies. The project team will adopt an inter-disciplinary research methodology combining distributed/parallel trust algorithm design, data-driven mathematical optimisation, modelling smart mobile users and easy to use financial transaction applications to address the research objectives.
The project outcomes aim to provide research-led disruptive innovation which would enable customised distributed fintech infrastructure provision and would lead to better designed “bespoke” fintech solutions. It is also expected to achieve significant economic impacts for the finance industry through mobile applications efficiently and effectively.
Improving performance of Byzantine fault tolerant algorithm in distributed trusted systems
Distributed trusted systems emulate distributed “trusted” services through a set of fault tolerant distributed algorithms, security protocols and business services, connected over the Internet. Particularly, the distributed ledger technology becomes disruptive and being adopted in industry, which has attracted much attention beyond the finance and banking sectors. While Byzantine fault tolerance (BFT) through consensus algorithms has proven to be reliable and robust for decentralised permissioned transaction systems with resilient to arbitrary failures. However, due to the adoption of BFT, such a system generally has poor transaction performance and scalability. This project will address such a widely recognised research challenge and will explore new parallel design of BFT algorithms, which combines with the data-driven multiple-objective and stochastic optimisation methods to compute numerical optimal solutions to maximise of consensus throughout.The project is led by Dr. Tiejun Ma and partnered with Huawei Edinburgh Lab. The research team aims to design and deliver a next generation distributed BFT framework solution with trustworthy decision-making and transactions while achieving high-speed performance as well as scalable to large-scale systems. The project outputs would be adopted beyond distributed transaction systems, and can be used widely for data management systems, cloud data systems, data centres, mobile, 5G and IoT services.