An analysis of numerical issues in neural training by pseudoinversion

R. Cancelliere, R. Deluca, M. Gai +2

Some novel strategies have recently been proposed for single hidden layer neural network training that set randomly the weights from input to hidden layer, while weights from hidde…