288 pp. per issue
6 x 9, illustrated
2014 Impact factor:

Neural Computation

Fall 1991, Vol. 3, No. 3, Pages 386-401
(doi: 10.1162/neco.1991.3.3.386)
© 1991 Massachusetts Institute of Technology
The Transition to Perfect Generalization in Perceptrons
Article PDF (859.29 KB)

Several recent papers (Gardner and Derrida 1989; Györgyi 1990; Sompolinsky et al. 1990) have found, using methods of statistical physics, that a transition to perfect generalization occurs in training a simple perceptron whose weights can only take values ±1. We give a rigorous proof of such a phenomena. That is, we show, for α = 2.0821, that if at least αn examples are drawn from the uniform distribution on {+1, −1}n and classified according to a target perceptron wt ∈ {+1, −1}n as positive or negative according to whether wt·x is nonnegative or negative, then the probability is 2−(√n) that there is any other such perceptron consistent with the examples. Numerical results indicate further that perfect generalization holds for α as low as 1.5.