Prof. Trevor MartinUniversity of Bristol, UK
The need for approximation in intelligent systems, or how to avoid "2+2 is approximately 5"
We are undergoing a revolution in artificial intelligence, driven by simultaneous increases in the computing power available and in the quantity of data that forms the input to algorithms. We can distinguish two distinct trends in this revolution. The first is taking us towards fully autonomous intelligent systems such as autonomic networks, driverless cars and the killer robots beloved by sci fi writers and the popular media. The second strand involves collaborative intelligent systems, where the complementary strengths of humans and machines work in partnership. In both cases, there is a gap between the crisp, binary representations used by computers and the more approximate, vague definitions used by humans. I will argue that this gap can be bridged by graded (fuzzy) representations, which give a mathematical framework to model the approximate hierarchical terms used in human language, as well as enabling management of the uncertainties, reliability and granularity inherent in much of the data.
Trevor Martin is Professor of Artificial Intelligence at the University of Bristol, UK and a BT Senior Research Fellow, working with the Security Futures Practice. His research covers soft computing in artificial intelligence applied to areas such as security analytics, extraction and integration of semi-structured information, soft concept hierarchies, and fundamental approaches to fuzzy uncertainty. In addition to substantial funding from BT, this work has been supported by the European Commission, MOD, GCHQ, EPSRC and DTI. He is a member of the editorial boards of journals such as Fuzzy Sets and Systems and Evolving Systems, and has served on many conference programme and organising committees, including IEEE Fuzzy Systems programme chair in 2007 and technical co-chair in 2010 and 2015. He is a co-organiser of the URSW (Uncertain Reasoning for the Semantic Web) series of workshops, chairs the IEEE Computational Intelligence Society’s Semantic Web Task Force and is a member of the IEEE's recently established FML (Fuzzy Markup Language) Standards group.
Back to Keynote Speakers