Log in
Results of research at Royal Holloway in Machine Learning — Support Vector Machine (SVM), kernel methods and conformal-prediction techniques — are at the source of the analytics and `Big Data' revolution, whose impact is transforming the economy (and society), from data mining to machine vision, from Web search to spam detection.
Although SVMs were not invented at Royal Holloway, our contributions include the original reference monograph (Vapnik 1998), basic underpinning theory (such as [3.2]), the first widely distributed open-source implementation [3.1], the first accessible textbook (Cristiniani and Shawe-Taylor 2000), and multiple extensions (such as [3.3] and [3.4]).
The delivery of interchangeable services across a range of educational platforms has been a long-term problem in the field of technology enhanced learning. The Institute for Educational Cybernetics (IEC) identified widgets as having a potential role in resolving this problem, and developed a widget server, Wookie, to provide a research tool to investigate this. The research is summarised in [4] and [6]. The work attracted international attention, and the server has been reused in a number of other projects to provide interoperable services, both in education and beyond, and including a number of European funded initiatives. The impact of the work was recognised and enhanced by its acceptance by the Apache Software Foundation as an incubator project. It has now graduated as Apache Wookie, and is a full Apache project.
Case Based Reasoning (CBR) is well suited to decision support in weak theory domains where important influences and interactions are not well understood. CBR retrieves and reuses similar cases that capture previous decisions, without reasoning about why/how the decision was made. Research at RGU has developed introspective learning technologies to capture knowledge that provides effective case retrieval and reuse in case-based systems. This self-optimised introspective CBR is embedded in a significantly changed process for insurance underwriting at Genworth Financials. Self-optimising retrieval selects relevant cases from Genworth's library of previous insurance cases, to be reused to assist decision-making of underwriters. The manual underwriting process is improved by increasing the consistency of underwriting decisions. Furthermore a 40% improvement in productivity is achieved for handling new insurance customers.
Professor Zhongyu (Joan) Lu's research contributed significantly to the development of a next-generation student response system (SRS) that is fully integrated with web services, Smartphones, multimedia and other ubiquitous technologies. By incorporating the use of widely available online equipment, the system has made SRS more affordable, easier to employ and applicable in a range of settings far more diverse than the traditional classroom scenario. It is now used in Europe and the US by both academia and industry and has served as the basis for a number of dedicated prototypes. Its success has also led to additional major funding streams for further research.
Impact is primarily economic and organizational, resulting from more effective leadership processes and practices by small firm owner-managers. The mechanism of impact was a programme known as LEAD (leadership, enterprise and development), which drew a significant community of owner-managers of smaller firms in Greater Merseyside into the Management School, to enable them to use research findings about managerial and entrepreneurial learning, leadership and business support in the running of their firms. The resulting impacts were on management practices and processes, and firm performances. Practitioners engaging with the University of Liverpool Management School (ULMS) LEAD programme experienced turnover increases averaging 21%. The beneficiaries are small firms, their employees and business support partnerships.
This case study describes impact arising from research into designing constructionist tools that provide personalisation, support and guidance to learners and teachers, resulting in software used in several schools, FE colleges and universities world-wide. Constructionist learning is founded on the principle of constructionism which argues for the pedagogical importance of building artefacts as a way of building mental representations. A key computational challenge in the design of tools that foster constructionist learning is to provide intelligent support that guides users towards productive interaction with the tool without constraining its creative potential.
Our research into learning through digital technologies has increased the focus on the importance of learning processes and context. The research developed new models of strategic evaluation and learning framework analyses as well as a new concept of MEGAcognition. These have shaped the development, customisation and implementation of more appropriate digital educational resources, nationally and internationally. Our research has involved and influenced key national and international companies and groups. Its users have been policy makers and developers, as well as teachers and pupils in primary and secondary schools. The research has: 1) influenced policy and practice developments nationally and internationally (in UK government departments and the e-strategy agency, and in five major resource development companies and corporations with international reach); 2) increased awareness of and engagement in learning opportunities (in four local authorities); 3) built capacity (in three resource development companies and projects); 4) offered insights into ways to develop, refine and customise educational products for specific audiences (in six resource development companies and local authorities); 5) raised awareness and understanding of educational concepts to non-academic audiences nationally and internationally (through 35 public and private seminars and keynote sessions to national and international audiences); 6) raised awareness of learning and pedagogical practices (in six major resource development companies and corporations).
Pioneering research into Inductive Logic Programming in the UOA led to the creation of Secerno Ltd. From 2008 Secerno attracted investment of approximately $20m and successfully released several updated versions of its product DataWall, based on this Oxford research. In May 2010 Oracle Corporation bought Secerno specifically to gain access to this technology, which now forms a core part of Oracle's database protection and compliance products. Oracle continues to develop the software, which is used across the globe by public entities and private companies to protect databases from internal and external attack and to ensure that they comply with relevant legislation. Customers include major businesses such as T-Mobile, which uses Database Firewall to protect 35 million users.
The Airbus company has used OntoREM, a semi-automated methodology developed at UWE Bristol, for developing systems' requirements specifications and improving the quality of such specifications. This has saved Airbus [text removed for publication] cost and time to develop aircraft operability requirements for wing design and industrialisation in two different aircraft programmes — with a significant increase in requirements reusability. It has enabled improved assessment of risk in advance of a project's start through prior estimation of the cost and time of developing requirements. This has allowed reliable forecasts and scheduling, and better management of the expectations of a project's key stakeholders.
Embedded software in the transportation sector (railway, automotive and avionics) needs to meet high reliability requirements because errors may have severe consequences. Research since 2008 in the UoA has developed effective reasoning technology to provide assurance that key error types are eliminated from embedded software, and has created novel algorithms to prove its integrity. Major players such as [text removed for publication] GM and Airbus have used technology developed in the UoA to verify the absence of errors. A particular advantage of this technology is its ability to reason about floating-point arithmetic, meaning that a much wider class of properties can be verified. The technology is widely distributed via third party operating systems and tool-sets.