Statistical physics of interacting neural networks

Wolfgang Kinzel; Richard Metzler; Ido Kanter

doi:10.1016/s0378-4371(01)00443-5

Statistical physics of interacting neural networks

Wolfgang Kinzel, Richard Metzler, Ido Kanter

Department of Physics - at Bar-Ilan University

University of Würzburg

Research output: Contribution to journal › Conference article › peer-review

2 Scopus citations

Abstract

Recent results on the statistical physics of time series generation and prediction are presented. A neural network is trained on quasi-periodic and chaotic sequences and overlaps to the sequence generator as well as the prediction errors are calculated numerically. For each network there exists a sequence for which it completely fails to make predictions. Two interacting networks show a transition to perfect synchronization. A pool of interacting networks shows good coordination in the minority game - a model of competition in a closed market. Finally, as a demonstration, a perceptron predicts bit sequences produced by human beings.

Original language	English
Pages (from-to)	44-55
Number of pages	12
Journal	Physica A: Statistical Mechanics and its Applications
Volume	302
Issue number	1-4
DOIs	https://doi.org/10.1016/s0378-4371(01)00443-5
State	Published - 15 Dec 2001
Event	International Workshop on Frontiers in the Physics of Complex Systems - Ramat-Gan, Israel Duration: 25 Mar 2001 → 28 Mar 2001

Bibliographical note

Funding Information:
The authors thank the Minerva Center and the German-Israel Science Foundation for support.

Funding

The authors thank the Minerva Center and the German-Israel Science Foundation for support.

Funders
German-Israel Science Foundation
Minerva Center

Access to Document

10.1016/s0378-4371(01)00443-5

Cite this

@article{c8c9012fe42146a7aa8e27f1747a5948,

title = "Statistical physics of interacting neural networks",

abstract = "Recent results on the statistical physics of time series generation and prediction are presented. A neural network is trained on quasi-periodic and chaotic sequences and overlaps to the sequence generator as well as the prediction errors are calculated numerically. For each network there exists a sequence for which it completely fails to make predictions. Two interacting networks show a transition to perfect synchronization. A pool of interacting networks shows good coordination in the minority game - a model of competition in a closed market. Finally, as a demonstration, a perceptron predicts bit sequences produced by human beings.",

author = "Wolfgang Kinzel and Richard Metzler and Ido Kanter",

note = "Funding Information: The authors thank the Minerva Center and the German-Israel Science Foundation for support.; International Workshop on Frontiers in the Physics of Complex Systems ; Conference date: 25-03-2001 Through 28-03-2001",

year = "2001",

month = dec,

day = "15",

doi = "10.1016/s0378-4371(01)00443-5",

language = "אנגלית",

volume = "302",

pages = "44--55",

journal = "Physica A: Statistical Mechanics and its Applications",

issn = "0378-4371",

publisher = "Elsevier B.V.",

number = "1-4",

}

TY - JOUR

T1 - Statistical physics of interacting neural networks

AU - Kinzel, Wolfgang

AU - Metzler, Richard

AU - Kanter, Ido

N1 - Funding Information: The authors thank the Minerva Center and the German-Israel Science Foundation for support.

PY - 2001/12/15

Y1 - 2001/12/15

N2 - Recent results on the statistical physics of time series generation and prediction are presented. A neural network is trained on quasi-periodic and chaotic sequences and overlaps to the sequence generator as well as the prediction errors are calculated numerically. For each network there exists a sequence for which it completely fails to make predictions. Two interacting networks show a transition to perfect synchronization. A pool of interacting networks shows good coordination in the minority game - a model of competition in a closed market. Finally, as a demonstration, a perceptron predicts bit sequences produced by human beings.

AB - Recent results on the statistical physics of time series generation and prediction are presented. A neural network is trained on quasi-periodic and chaotic sequences and overlaps to the sequence generator as well as the prediction errors are calculated numerically. For each network there exists a sequence for which it completely fails to make predictions. Two interacting networks show a transition to perfect synchronization. A pool of interacting networks shows good coordination in the minority game - a model of competition in a closed market. Finally, as a demonstration, a perceptron predicts bit sequences produced by human beings.

UR - https://www.scopus.com/pages/publications/0035894725

U2 - 10.1016/s0378-4371(01)00443-5

DO - 10.1016/s0378-4371(01)00443-5

M3 - ???researchoutput.researchoutputtypes.contributiontojournal.conferencearticle???

AN - SCOPUS:0035894725

SN - 0378-4371

VL - 302

SP - 44

EP - 55

JO - Physica A: Statistical Mechanics and its Applications

JF - Physica A: Statistical Mechanics and its Applications

IS - 1-4

T2 - International Workshop on Frontiers in the Physics of Complex Systems

Y2 - 25 March 2001 through 28 March 2001

ER -

Statistical physics of interacting neural networks

Abstract

Bibliographical note

Funding

Access to Document

Other files and links

Fingerprint

Cite this