Face recognition using a digital neural network with self-organising capabilities
Michael Kerin, T.J. Stonham · 2002
An approach to face recognition using neural networks is presented. All automatic training procedure based on self-organization is defined. This procedure partitions the data into disjoint classes containing recognizable attributes and can be used to replace a human operator in selecting valid data for training purposes. The nodal operation within the neural network was initially based on combinational logic functions and has been developed to take account of the dominance of salient features in the data. Experimental results demonstrate the system's facial recognition capabilities using multiple classes of data.>