Ehresmann connections and feedforward neural networks

被引：0

作者：

Pearson, DW ^{[1
]}

机构：

[1] EMA, LGI2P, Nonlinear & Uncertain Syst Grp, F-30035 Nimes 1, France

来源：

MATHEMATICAL AND COMPUTER MODELLING | 1999年 / 29卷 / 09期

关键词：

Ehresmann connections; feedforward neural networks; vertical vector fields;

D O I：

10.1016/S0895-7177(99)00078-3

中图分类号：

TP39 [计算机的应用];

学科分类号：

081203 ; 0835 ;

摘要：

In this paper, we describe how Ehresmann connections can be used to study certain properties of feedforward neural networks. Essentially, we calculate a Lie group approximation to the structure of the inverse image set above a certain point in the output space and this structure can then be locally transported to the inverse image above a, neighbouring point in the output space by means of an Ehresmann connection. This enables us to find a continuous approximation to the underlying topological structure of the data from discrete data pairs (input/output pairs). (C) 1999 Elsevier Science Ltd. All rights reserved.

引用

页码：17 / 25

页数：9

共 7 条

[1] A THEOREM ON HOLONOMY [J].

AMBROSE, W ;

SINGER, IM .

TRANSACTIONS OF THE AMERICAN MATHEMATICAL SOCIETY, 1953, 75 (NOV) :428-443

[2]

Ehresmann C., 1950, C TOP BRUX, P29

[3]

Golub GH, 2013, Matrix Computations, V4

[4]

HERMANN R, 1964, CONTRIB DIFF EQ, V3, P199

[5] Approximating vertical vector fields for feedforward neural networks [J].

Pearson, DW .

APPLIED MATHEMATICS LETTERS, 1996, 9 (02) :61-64

[6] LEARNING REPRESENTATIONS BY BACK-PROPAGATING ERRORS [J].

RUMELHART, DE ;

HINTON, GE ;

WILLIAMS, RJ .

NATURE, 1986, 323 (6088) :533-536

[7]

STEENROD N, 1951, TOPOLOGY FIBRE BUNDL

← 1 →