1 paper · 1 filter
H. C. Rae, P. Sollich, A. C. C. Coolen
We solve the dynamics of on-line Hebbian learning in large perceptrons exactly, for the regime where the size of the training set scales linearly with the number of inputs. We cons…