Neural Computing

Lecturer: Dr J.G. Daugman (jgd1000@cl.cam.ac.uk)

No. of lectures: 12

Prerequisite course: Continuous Mathematics  

Natural versus artificial substrates of intelligence.

Investigation into how biological nervous systems accomplish many of the goals of machine intelligence, but using radically different strategies, architectures, and hardware. Comparison of those differences, and examination of their importance or irrelevance. Levels of analysis; mechanism and explanation; philosophical issues.

Neurobiological wetware, processing and signalling.

Human brain architecture. Sensation and perception. Learning and memory. Information content of neural signals. Spike generation processes. Stochastic neural codes; Principal Components Analysis of spike trains. Visual operators that encode, analyse, and represent image structure. Face recognition, perceptual learning, and higher cognitive functions. How stochastic communications media, unreliable and randomly distributed hardware, very slow and asynchronous clocking, and imprecise connectivity, yield unrivalled performance in real-time tasks involving perception, learning, and motor control.

Artificial neural networks and neurocomputing.

Simplifying models that seek to emulate biological designs for computing via connectivity dynamics and statistics, instead of by formal sequential algorithms. Supervised and unsupervised learning by adjustment of network connectivity. Synapses and update rules. Layered networks; feedforward, backpropagation, and counterpropagation. Hebbian synapses; Hopfield and Widrow-Hoff networks. Self-organizing networks. Competitive and cooperative networks. Energy minimization, relaxation, regularization. Adaptive resonance and chaos. Genetic algorithms.

Recommended books:

Haykin, S. (1994). Neural Networks: A Comprehensive Foundation. MacMillan.

Hecht-Nielsen, R. (1991). Neurocomputing. Addison-Wesley.