We are honoured to share that Dr Dalton is one of the fifteen recipients of the prestigious Turing AI Acceleration fellowship. It is a five year fellowship from UKRI in collaboration with Alan Turing Institute on “Neural Conversational Information Assistants”. See the official AI Fellowship Announcement. Associated with the Fellowship there are multiple PhD positions as well as post-doctoral research associate opportunities that will be available in the GRILL Lab.
For context: “The Turing AI Acceleration Fellowships will support some of our leading researchers to progress their careers and develop ground-breaking AI technologies with societal impact. By enhancing collaboration between academia and industry and accelerating these transformative technologies these Fellowships will help to maintain and build on the UK’s position as a world leader in AI.”
Our team’s research on neural deep-learning models for conversational assistants will make significant advances to language understanding by bridging search-based information agents and symbolic entity-based language understanding. We propose new methods for informational assistants: scaling open-domain Transformer architectures to longer sequences using distilled word and entity vector representations, bridging the ‘structure chasm’ by automatically learning task schemas using weak supervision from text and implicit behaviour data, and using deep reinforcement learning and simulation approaches to develop effective interaction policies.
It will enable an assistant to have in-depth discussions on key topics. For example, in a conversation with an emerging climate activist, it could interact to determine the reason for the research, ask questions to determine their knowledge and interests, and assess their ‘climate literacy’. It will engage in a collaborative information dialogue to find and summarize the most important information – from playing excerpts of Greta Thunberg’s latest speech or perhaps buying Sir David Attenborough’s latest book. It will understand key topics and assist in taking actions, for example by signing up to take part in events around the upcoming COP26 summit in Glasgow in 2021 or walk through switching to the best ‘green’ energy provider. And further, our research will enable conversational systems to remember key topics, events, and actions that may span days or even months.
The project includes collaborations with key industry and third-sector partners including the BBC Scotland Voice and AI group, PolyAI, MoodAgent, HarperCollins publishing, and others. If you are interested in opportunities for collaboration, please reach out.
If you are interested in opportunities to work with us, please reach out and leave your details.