Research areas
Current research is with the Intelligent Computation Group at the University of St Andrews.
Why are people still smarter than digital computers? A serial digital computer can perform millions of operations per second, while the human brain has a firing cycle for each neuron of the order of milliseconds. One conclusion from these facts is that the human brain's superior abilities are made possible through massive parallel distributed processing and representation. The incorporation of such parallelism into artificial systems with semi-conductor switching speeds offers the potential of very powerful capabilities. For the capabilities to be realised as actual abilities though, the parallelism has to be organised appropriately. Artificial neural networks are an attempt to provide systems incorporating these features effectively by modelling abstract computational features of the brain's information processing.
The aim of the Intelligent Computation Group at St Andrews is to provide such effective computation through modelling of both biological and cognitive insights into intelligent information processing and representation. This modelling also incorporates other artificial intelligence techniques such as Symbolic AI and Artificial Life into neural systems where they are appropriate. Heuristic search is used to provide benchmarks for the effectiveness of the computation.
Selected publications
-
High Quality Goal Connection For Nonholonomic Obstacle Navigation Allowing For Drift Using Dynamic Potential Fields
Weir, M. K. & Bott, M. P., 2010, 2010 IEEE International Conference on Robotics and Automation (ICRA). IEEE, p. 3221-3226 6 p.Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
-
Enabling nonholonomic smoothness generically allowing for unpredictable drift
Weir, M. K., Lewis, J. P. & Bott, M. P., 2008, 10th International Conference on Control, Automation, Robotics and Vision, 2008. ICARCV 2008. IEEE, p. 2072-2077 6 p.Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
-
POTBUG: A mind's eye approach to providing BUG-like guarantees for adaptive obstacle navigation using dynamic potential fields
Weir, M., Buck, A. & Lewis, J., 2006, From Animals to Animats 9: 9th International Conference on Simulation of Adaptive Behaviour. Nolfi, S., Baldassarre, G., Calabretta, R., Hallam, J., Marocco, D., Meyer, J., Miglino, O. & Parisi, O. (eds.). Springer-Verlag, p. 239-250 12 p. (Lecture Notes in Computer Science; vol. 4095).Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
-
Agent Navigation Using Potential Fields and Forward Chaining
Bell, G. & Weir, M. K., 2004.Research output: Contribution to conference › Paper
-
Forward Chaining for Robot and Agent Navigation using Potential Fields
Bell, G., Weir, M. K. & Estivill-Castro, V., 2004, p. 10. 10 p.Research output: Contribution to conference › Paper
-
An Approach to Guaranteeing Generalisation in Neural Networks
Weir, M. K. & Polhill, J. G., Oct 2001, In: Neural Networks. 14, 8, p. 1035-1048 14 p.Research output: Contribution to journal › Article › peer-review
-
Neural Steering: Difficult and Impossible Sequential Problems for Gradient Descent
Milligan, G., Weir, M. K., Lewis, J. P., Mira, J. & Prieto, A., Jun 2001, p. 8. 8 p.Research output: Contribution to conference › Paper
-
Using Tangent Hyperplanes to Direct Neural Training
Weir, M. K., Lewis, J., Milligan, G., Bothe, H. & Rojas, R., 2000.Research output: Contribution to conference › Paper
-
Improving Generalisation Using Neural Bi-Directional Convergence
Weir, M. K., 1999.Research output: Contribution to conference › Paper
-
Subgoal Chaining and the Local Minimum Problem
Weir, M. K. & Lewis, J., 1999, p. 1844-1849.Research output: Contribution to conference › Paper