After completing his PhD in Theoretical Physics, Dr Bullinaria worked on superstring theory and quantum gravity at the University of Durham. He then took a career break and retrained in Artificial Intelligence. Since then he worked on neural network models of brain damage at the University of Edinburgh, on neural network models of language acquisition at Birkbeck College London, the statistical analysis of large text corpora at Macquarie University Sydney, and on computational models of oculomotor control and the importance of evolutionary factors in developmental models at the University of Reading. He moved to the University of Birmingham in March 2001.
Dr Bullinaria is a Senior Lecturer in Computer Science and also a Deputy Director of Cercia. His research activities are currently in the broad area of natural computation, but particularly in the modelling of neural and evolutionary systems. Applications range from brain modelling (such as neuropsychology, language processing, motor and sensor control systems) to artificial systems building (such as novel neural network architectures and learning algorithms, and robot controllers). Over the last few years he has become particularly interested in using evolutionary computation techniques to generate more powerful and reliable neural network systems than can be formulated by hand. This approach can address many known problems in setting parameters (such as age dependent learning rates) and architectures (including modularity and sub-system interfaces) significantly better than human researchers can. In addition, these techniques can be employed directly in a wide range of applied/industrial areas.
John Bullinaria's personal home page
