One-pass training of optimal architecture auto-associative neural network for detecting ectopic beats
Clifford G., Tarassenko L., Townsend N.
A method for detection of ectopic beat in the electrocardiogram using an auto-associative neural network was presented. It utilises principal component analysis for optimizing the complexity of the neural network and uses singular value decomposition for determining the initial values for the weights. The trained weights could be updated on-line with back propagation with floating point operations.