Finite size scaling in neural networks
Walter Nadler, Wolfgang Fink
Abstract
We demonstrate that the fraction of pattern sets that can be stored in single- and hidden-layer perceptrons exhibits finite size scaling. This feature allows to estimate the critical storage capacity αc from simulations of relatively small systems. We illustrate this approach by determining αc, together with the finite size scaling exponent ν, for storing Gaussian patterns in committee and parity machines with binary couplings and up to K=5 hidden units.
Create a lesson
Related papers
Low-temperature magnetism and spin dynamics in the disordered triangular-lattice Yb3+ compound LiCaYb5(BO3)6
Monika Jawale, Saikat Nandi, Prashanta K. Mukharjee et al.
Neural Renormalization Group Flow for Percolation
Anaclara Alvez, Luca Camagna, Sergio Chibbaro et al.
Dynamical phase selection controls compute scaling in looped transformers
Gunn Kim
Semi-localized ground state in a 1D system with long-range hopping
Murod S. Bahovadinov, Faridun N. Jalolov, Vladimir E. Kravtsov et al.
Defect states in three-dimensional diamond photonic band gap crystals
Julia Rocha, Bart A. van Tiggelen, Ad Lagendijk et al.
Disorder-induced conducting edges on Kagomé lattice
A. Chmeruk, D. Jones, L. Chioncel