Permutation parity machines for neural synchronization

被引：8

作者：

Reyes, O. M. ^{[1
,2
]}

Kopitzke, I. ^{[1
]}

Zimmermann, K-H ^{[1
]}

机构：

[1] Hamburg Univ Technol, Inst Comp Technol, D-21071 Hamburg, Germany

[2] Univ Ind Santander, Escuela Ingn Elect Elect & Telecomunicac, Bucaramanga, Colombia

来源：

JOURNAL OF PHYSICS A-MATHEMATICAL AND THEORETICAL | 2009年 / 42卷 / 19期

关键词：

D O I：

10.1088/1751-8113/42/19/195002

中图分类号：

O4 [物理学];

学科分类号：

0702 ;

摘要：

Synchronization of neural networks has been studied in recent years as an alternative to cryptographic applications such as the realization of symmetric key exchange protocols. This paper presents a first view of the so-called permutation parity machine, an artificial neural network proposed as a binary variant of the tree parity machine. The dynamics of the synchronization process by mutual learning between permutation parity machines is analytically studied and the results are compared with those of tree parity machines. It will turn out that for neural synchronization, permutation parity machines form a viable alternative to tree parity machines.

引用

页数：20

共 16 条

[1]

[Anonymous], 2004, Statistical Mechanics of Learning

[2]

[Anonymous], THESIS HAMBURG U TEC

[3]

[Anonymous], 1998, NEURAL NETWORKS

[4]

[Anonymous], 1983, FINITE MARKOV CHAINS

[5]

[Anonymous], 1980, Finite Markov Processes and their Applications

[6]

[Anonymous], An Introduction to Probability Theory and its Applications

[7]

[Anonymous], THESIS HAMBURG U TEC

[8]

[Anonymous], 1999, Mathematical Statistics

[9]

Ditto W, 2002, NATURE, V415, P736, DOI 10.1038/415736b

[10] Secure exchange of information by synchronization of neural networks [J].

Kanter, I ;

Kinzel, W ;

Kanter, E .

EUROPHYSICS LETTERS, 2002, 57 (01) :141-147

← 1 2 →