Applications of AI
- Creating Machine Intelligence with Intelligent Interactive Visualisation collaboration with Montvieux Ltd.
This collaborative research project aims to resolve three of the challenges to building a predictive event analysis system:
- the encoding of event data in a format suitable for input to a neural network
- the training of a neural network to forecast events
- the use of introspection techniques to identify learned patterns and to explain predictions.
- Hypergraph Partitioning with Memetic Algorithms
Investigating novel methods for impoving the way that huge problems can be sub-divided to make them more tractable for optimisation.
- Improved Search Methods for Improving Query Relevance and Accuracy Knowledge Transfer Partnership with Multicomm Ltd./ Paxport Group UK 2016-17
This project developed novel combinations of Machine Learning in Recommender Systems tools to improve the relevance of the results returned by the company's' Find-and-Book service, which is used by travel agents worldwide.
- Conversational AI Agent for Virtual Air Traffic Control (vATC) 2014-15. Innovation for Growth Project with Micro Nav Ltd.
Adapted and improved the speech recognition engines and developed new methods for creating targetted training scripts.
- Extensões industriais ao problema de planejamento e programação da produção.
Funded by Fundação Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), Ministry of Education, Brazil,Oct. 2009 to Oct. 2011. Co-Investigator.
Financed a month-long visit to two universities in Brazil for collaborative research.
- Industrial Extensions to Production Planning and Scheduling EU FP7 "Marie Curie International Research Staff Exchange Scheme (IRSES)" , with partner universities from Portugal and Brazil ( Faculdade de Engenharia da Universidade do Porto, Universidade de Sao Paulo, Universidade Federal de Sao Carlos, Universidade Estadual Paulista - UNESP)March 2010 - Feb 2013
This project concerns the development of improved techniques to help industries improve efficiency in areas such as lot-sizing and scheduling, cutting and packing.
- Evolutionary Algorithms for Computational Biology. Research project funded by Sandia National Laboratories, New Mexico, June-July2001.
- Investigating Applications of AI to create more realistic flight simulation exercises January-March 2013: Micro-electronics iNet project with Micro Nav Ltd.
- Microprocessor Design Verification Using Genetic Algorithms: Applications to LSU verification. Research project for STMicrolectronics plc. 1995.
- Microprocessor Design Verification Using Genetic Algorithms: A Feasibility Study. Research project for SGS-Thomson Microelectronics plc. 1995.
- The Automatic Generation of Software Test Data using Genetic Algorithms. Research project for SGS-Thomson Microelectronics plc. 1994.
Statistical Disclosure Control
- Application of Meta-heuristics for Statistical Disclosure Control of Business Data” Office for National Statistics, September 2013 – January 2014.
Embedded UWE-developed tools within ONS and develped new methods for partitioning huge data sets.
- Mathematical Modelling of Statistical Disclosure Control. EPSRC Mathematical CASE award with UK Office for National Statistics, 2007-2011.
- Evaluation of Heuristic Approaches to Statistical Disclosure Control. Consultancy for Office for National Statistics. Feb-April 2007.
- Improvements to Cell Suppression in Statistical Disclosure Control. Research project funded by Office of National Statistics, Dec 2005-March 2006. Joint with Alistair Clark
These research collaborations with the UK Office of National Statistics investigate the use of Evolutionary Algorithms for
Statistical Disclosure Control.
This work concerns how to protect confidentiality in publicly available datasets (e.g. census, industrial outputs) while maximising
the usefulness of what is published. See here for a user-friendly discussion in the context of UWE's "Research with Impact" agenda.
- Interactive Evolutionary Web-Page Customisation EU funded "Knowledge Escalator" project with CX Partners and Pure Useability 2009
- DynaVis: Dynamically Reconfigurable Quality Control for Manufacturing and Production Processes Using Learning Machine Vision. EU FP6 STRP 016429, 2005-2009.
Applied Machine Learning techniques to create dynamically reconfigurable machine vision systems for manufacturing quality control.
As well as developing classification systems, my team particularly looking at "human" aspects which affect the
training provided to the machines - such as differences
between operators, and the effects of fatigue and lack of
confidence. We also developed systems that gave feedback to the users
as to how well the systems were learning under their guidance, and how
well they might be expected to learn from further input.
Outreach and Pedagogical Research