Previous Research projects

Previous Research projects within AIAI

MathSoMac: the social machine of mathematics 2018-2023 (Extension)
July 2018 - Jan 2023  Ursula Martin

This project is an extension of Ursula Martin's EPSRC Established Career Fellowship ‘MathSoMac: the social machine of mathematics’ (2014-2018), and has two main themes:

Beyond Inference: in collaboration with IMKT, the University of Waterloo (Stephen Watt) and the University of Pittsburgh (Tom Hales), we look at the variety of ways which mathematical statements are used, in particular in contexts other than proof, focussing on Tom Hales’s development of “formalised abstracts” as a bridge between formal and informal mathematics. Towards impact: in collaboration with Elsevier and major museums, looks at the impact and cultural capital of foundational research, building on a major study of historical archive material in the twentieth century history of computing. A particular interest is the history and contribution of women in computing.

UNBIAS: Emancipating Users Against Algorithmic Biases for a Trusted Digital Economy
September 2016 - August 2018    Michael Rovatsos

UnBias £1.1-million research project funded by EPSRC under the Trust, Identity, Privacy and Security (TIPS) programme.

The project is a collaboration between researchers the University of Nottingham, the University of Oxford, and CISA at the University of Edinburgh. Since 2016 we have been looking at the user experience of algorithm driven internet services and the process of algorithm design. A large part of this work are user group studies to understand the concerns and perspectives of citizens. UnBias aims to provide policy recommendations, ethical guidelines and a ʽfairness toolkitʼ co-produced with young people and other stakeholders that will include educational materials and resources to support understanding about online environments as well as raise awareness among online providers about the concerns and rights of young internet users.

Within the project, CISA researchers focus on developing fairer data-driven algorithms and tools for evaluating these.

A Query Answering Framework Using Functional Inferences Over heterogeneous Data
October 2016 - January 2019    Alan Bundy
The internet provides a vast store of information of mixed quality and many different formats. We not only want to retrieve already available information from it, but to infer new information. This includes prediction, e.g., "what will be the UK's population in 2025", which can be achieved by creating a graph from census data and then extrapolating it forward in time. In his PhD work, Kwabena (Kobby) Nuamah created the RIF system, which breaks a query into sub-queries, finds the required sub-answers on the internet, curates them into a common format and then combines the sub-answers into an answer to the original query, with error bars estimated from the reliability of the sources and the inference methods. RIF will be adapted to and evaluated on various queries that are interesting to Huawei and its customers.
May 2015 - April 2019    Malcolm Atkinson
Towards Explainable and Robust Statistical AI: A Symbolic Approach
May 2018 - July 2019    Vaishak Belle

Data science provides many opportunities to improve private and public life, and it has enjoyed significant investment in the UK, EU and elsewhere. Discovering patterns and structures in large troves of data in an automated manner — that is, machine learning — is a core component of data science. But a deep challenge here is this: when can we convincingly deploy these methods in our workplaces? For example:  (a) how can we elicit intuitive explanations from the decisions offered by these methods? (b) would these decisions be amenable to suggestions/preferences from non-expert users? (c) how can we provide correctness guarantees for the underlying computations?

The EPSRC project " Towards Robust Statistical AI: A Symbolic Approach" attempts first steps towards this direction. 

Digiflow: Digitizing Industrial Workflows
January 2018 - December 2019    Jacques Fleuriot

Digiflow is an Industry 4.0 project funded by EIT Digital under its  Digital Industry Action Line. It revolves around a combination of IoT sensors, cloud infrastructure and workflow technologies to enable manufacturing companies to better monitor their shop floors and provide decision support towards efficiency optimisation. Partners include CREATE-NET from the Fondazione Bruno Kessler (FBK) in Italy, technology innovation company Reply Santer, Italian SME ThinkINside, and the School of Informatics’ WorkflowFM team consisting of Jacques Fleuriot,  Petros Papapanagiotou and James Vaughan.

This project allows a unique combination of state-of-the-art technologies and skills brought together by each partner: (1) fog and cloud infrastructures for the management of IoT devices and the orchestration of associated processes by FBK, (2) expertise in product and business development for technological solutions with a worldwide reach by Reply Santer, (3) novel sensor technologies to track and analyze the activity of assets in a manufacturing floor by ThinkINside, and (4) the latest in workflow modelling and management research, including performance analytics, by the University of Edinburgh. The end-goal is a product that provides real-time monitoring, helps determine root causes of delays and inefficiencies, improves throughput, and reduces operational costs. The technology will be commercialised by Reply Santer. This is a key opportunity to make use of some of the latest research from the School of Informatics in a real, deployed solution that has the potential to create significant impact within manufacturing.

DARE: Delivering Agile Research Excellence on European E-Infrastructures
January 2018 - December 2020    Malcolm Atkinson



MathSoMac: the social machine of mathematics 2018-2023 (Extension)

MathSoMac: the social machine of mathematics 2018-2023 (outline)