Speech and Language Technologies (SLTs) are a central element of Artificial Intelligence (AI). Located at a world leading research institution in the field, this UKRI AI Centre for Doctoral Training (CDT) will host at least 60 PhD students over a period of 8 years.
AI had been identified in the UK Government’s Industrial Strategy and one of the four Grand Challenges and an area for strategic growth. The CDT will help address this by training students in the theory and application of computational speech and language processing.
The centre is hosted within the Department of Computer Science which has an international reputation for the quality of its research and teaching. In the 2014 Research Excellence Framework (REF – the UK Government’s national assessment of university research), 92% of our research work was rated world leading (4*) or internationally excellent (3*) in terms of its originality, significance and rigour.
The Department has a REF grade point average of 3.39, ranking them 5th out of 89 computer science departments in the UK. The department currently hosts 131 research students and at present it has research funding from many sources including from RCUK of over £14m.
The Department is a national leader and a highly respected global player in Speech and Language Technologies, with arguably the largest mass of Speech and Language Technologies researchers in a single department in the UK. The academic quality of the Speech and Language Technologies team is reflected in their very significant contribution the high score of the Department of Computer Science in the last REF exercise and by the number of EPSRC/ERC fellowships awarded to team members. The team has an outstanding track record in research grant awards, impact and award of PhDs; in addition, they have an extensive set of international industrial collaborators, across many sectors and ranging from SMEs to global players. In summary, they have the know-how, the experience, the external support and the passion to deliver outstanding research outputs in SLT with demonstrable national and international need.
EPSRC prosperity outcomes
The Engineering and Physical Sciences Research Council (EPSRC) have identified that the future competitiveness and creativity of the UK economy requires the successful development of world-leading products, processes and technology based on the discovery and innovation in the mathematical and physical sciences, information and computing technologies, and engineering.
The EPSRC aims to anticipate economic and social change, and re-skill the UK workforce with a particular requirement to achieve technical leadership through the development of future scientists, engineers and technologists.
The CDT is committed to contributing to the four EPSRC prosperity outcomes
The CDT targets SLT applications in the areas of improving healthcare (e.g., assistive technology, clinical applications of speech technology, analysing medical forums for newly reported drug side effects, problems with healthcare provision, misinformation).
SLT applications support a safe and trusted society (e.g., analysing social media for hate speech, political abuse, terrorist activities); supporting policy makers, government organisations, media, and political scientists in studying, e.g., barriers to inclusion, influence of alternative media on the democratic processes and society and online propaganda.
The CDT will include joint PhD projects with the Grantham Centre for Sustainable Futures, which will be aligned with ongoing joint research using NLP infrastructure.
The focus will be on SLT applications that boost productivity via robotics, IoT, and big data analytics. The CDT fits well within the “AI and Data-Driven Economy” grand challenge of the Government’s Industry Strategy Challenge Fund, which specifically mentions SLT.
Professor Thomas Hain
Head of the Speech and Hearing Research Group
Prof Hain is a world leader in speech recognition, heads the Voicebase Centre for Speech and Language Technology and is a leader in the speech community.
‘Talking and listening, understanding and expressive communication are skills that we all have. To this day we struggle to build machines that come close to human abilities. To explore and invent methods that allow us to recognise what is spoken, to understand, transform and interpret human communication has been the focus of my research. I am interested in machine learning methods that allow us to model communication and interaction, to be able to help people communicate, learn, and engage with new technology.’
Example of possible topics for supervision include; advanced modelling of speech processes, models of acoustic environments or of language, relationship between languages, and systems that transcribe spoken words, analyse them, transform the signal or the language, and on systems that respond to you and learn from you.
Professor Rob Gaizauskas
Head of the Natural Language Processing Research Group
Prof Gaizauskas is internationally known for his research on information extraction and text mining, temporal information processing, question answering and summarisation.
‘Can we build we build computer programs that “understand” human language? This question is of interest from both a cognitive science/linguistic perspective and from an applied/engineering perspective. What are the syntactic/semantic and pragmatic mechanisms available in human languages and how do intentional agents deploy them to communicate and accomplish goals in the world? How can we use our current, partial understanding of NLP to engineer applications that help people to gain better access to information in massive amounts of textual data and to dynamically interact with intelligent agents via NL dialogue?’
Example of possible topics for supervision include; information extraction/text mining; automatic summarization; semantic annotation of temporal and spatial information; automatic generation of image descriptions; common sense knowledge and NL understanding; task-oriented dialogue.
Ieso Digital Health
Jam Creative Studios
King’s College London