Anyone know how to train a perceptron with k-fold or hold-one-out testing?
Do you delete the known output to prevent the perceptron from memorising it? Or do you delete the perceptron's network output for the relevant testing set until it has been trained up?