TY - JOUR
T1 - Storage properties of correlated perceptrons
AU - Engel, A.
AU - Malzahn, D.
AU - Kanter, I.
PY - 1998/5
Y1 - 1998/5
N2 - Feedforward multilayer neural networks implementing random input—output mappings develop characteristic correlations between the activity of their hidden nodes which are important for the understanding of their storage and generalization performance. It is shown how these correlations can be calculated within the replica-symmetric approximation. Replacing the multilayer network by an ensemble of perceptrons displaying the same correlations the relative influence of these correlations on the storage capacity can be studied.
AB - Feedforward multilayer neural networks implementing random input—output mappings develop characteristic correlations between the activity of their hidden nodes which are important for the understanding of their storage and generalization performance. It is shown how these correlations can be calculated within the replica-symmetric approximation. Replacing the multilayer network by an ensemble of perceptrons displaying the same correlations the relative influence of these correlations on the storage capacity can be studied.
UR - http://www.scopus.com/inward/record.url?scp=0032072874&partnerID=8YFLogxK
U2 - 10.1080/13642819808205042
DO - 10.1080/13642819808205042
M3 - ???researchoutput.researchoutputtypes.contributiontojournal.article???
AN - SCOPUS:0032072874
SN - 1364-2812
VL - 77
SP - 1507
EP - 1513
JO - Philosophical Magazine B: Physics of Condensed Matter; Statistical Mechanics, Electronic, Optical and Magnetic Properties
JF - Philosophical Magazine B: Physics of Condensed Matter; Statistical Mechanics, Electronic, Optical and Magnetic Properties
IS - 5
ER -