Pattern Recognition & Neural Networks
Shift-invariant neural nets, CMAC classifiers, hand-printed character recognition, and position-independent pattern matching.
Emeritus Professor · Chartered Engineer · Founder
David Elliman
Background
I've worked in advanced technology industries or as an academic all my working life, often both at the same time. I started at Rolls-Royce, moved to Quest Automation, and then became an academic at the University of Nottingham, progressing to full professor and leading a successful research group in pattern recognition, neural networks, and intelligent systems.
After a stint on an interesting government AI project, I joined Monvieux Ltd — which had expanded to 100 staff by the time I left to follow a new initiative with Neuro-Symbolic Ltd. I was excited by the prospect of combining the safety of symbolic logic with the extraordinary, but unreliable, talent of large language models.
I'm interested in the development of machine learning and AI, in software engineering, and I have a side interest in quantum physics and electronics.
Research Interests
39 publications spanning pattern recognition, neural networks, handwriting and document analysis, ontologies, constraint-based scheduling, and AI agents. A few highlights:
Document Analysis & Handwriting
Arabic handwriting recognition with HMMs, graphics recognition, vectorization algorithms, and arrowhead detection.
Knowledge Representation
Ontology languages for the semantic web, XML data models, and automated document ontology derivation.
Constraint Solving & Optimisation
Genetic algorithm timetabling, graph colouring, reactive ant colony optimisation, and intelligent agent routing.
My physics work studies finite quantum registers, measurement records and explicitly supplied lattice actions. The physics papers distinguish mathematical results from assumptions needed for a physical interpretation.
Away from the Screen
Settled in NW Gloucestershire with Jane, Rufus the Cockerpoo and Grace the British Shorthair. When I'm not wrangling language models or lattice simulations, there's always plenty to do in the garden and garage.
Interested in Neuro-Symbolic AI?
Whether you're exploring automation for regulated workflows or curious about the Circlette Theory, I'd love to hear from you.