Analytical and numerical study of internal representations in multilayer neural networks with binary weights

Simona Cocco; Rémi Monasson; Riccardo Zecchina

doi:10.1103/PhysRevE.54.717

Analytical and numerical study of internal representations in multilayer neural networks with binary weights

Simona Cocco
, Rémi Monasson
, Riccardo Zecchina

Sezione di Roma
Center for Atomic-scale Materials Physics (CAMP)
Politecnico di Torino

Résultats de recherche: Contribution à un journal › Article › Revue par des pairs

Résumé

We study the weight space structure of the parity machine with binary weights by deriving the distribution of volumes associated to the internal representations of the learning examples. The learning behavior and the symmetry breaking transition are analyzed and the results are found to be in very good agreement with the extended numerical simulations.

langue originale	Anglais
Pages (de - à)	717-736
Nombre de pages	20
journal	Physical Review E
Volume	54
Numéro de publication	1
Les DOIs	https://doi.org/10.1103/PhysRevE.54.717
état	Publié - 1 janv. 1996
Modification externe	Oui

Accès au document

10.1103/PhysRevE.54.717

Contient cette citation

@article{593819a1784e462e863cdf21aedd7aff,

title = "Analytical and numerical study of internal representations in multilayer neural networks with binary weights",

abstract = "We study the weight space structure of the parity machine with binary weights by deriving the distribution of volumes associated to the internal representations of the learning examples. The learning behavior and the symmetry breaking transition are analyzed and the results are found to be in very good agreement with the extended numerical simulations.",

author = "Simona Cocco and R{\'e}mi Monasson and Riccardo Zecchina",

year = "1996",

month = jan,

day = "1",

doi = "10.1103/PhysRevE.54.717",

language = "English",

volume = "54",

pages = "717--736",

journal = "Physical Review E",

issn = "2470-0045",

number = "1",

}

TY - JOUR

T1 - Analytical and numerical study of internal representations in multilayer neural networks with binary weights

AU - Cocco, Simona

AU - Monasson, Rémi

AU - Zecchina, Riccardo

PY - 1996/1/1

Y1 - 1996/1/1

N2 - We study the weight space structure of the parity machine with binary weights by deriving the distribution of volumes associated to the internal representations of the learning examples. The learning behavior and the symmetry breaking transition are analyzed and the results are found to be in very good agreement with the extended numerical simulations.

AB - We study the weight space structure of the parity machine with binary weights by deriving the distribution of volumes associated to the internal representations of the learning examples. The learning behavior and the symmetry breaking transition are analyzed and the results are found to be in very good agreement with the extended numerical simulations.

U2 - 10.1103/PhysRevE.54.717

DO - 10.1103/PhysRevE.54.717

M3 - Article

AN - SCOPUS:0007132343

SN - 2470-0045

VL - 54

SP - 717

EP - 736

JO - Physical Review E

JF - Physical Review E

IS - 1

ER -

Analytical and numerical study of internal representations in multilayer neural networks with binary weights

Résumé

Accès au document

Empreinte digitale

Contient cette citation