Multifractal Analysis of the Coupling Space of Feed-Forward Neural Networks
arXiv:cond-mat/9512176 · doi:10.1103/PhysRevE.53.R2064
Abstract
Random input patterns induce a partition of the coupling space of feed-forward neural networks into different cells according to the generated output sequence. For the perceptron this partition forms a random multifractal for which the spectrum $f(α)$ can be calculated analytically using the replica trick. Phase transition in the multifractal spectrum correspond to the crossover from percolating to non-percolating cell sizes. Instabilities of negative moments are related to the VC-dimension.
10 pages, Latex, submitted to PRL