Small Depth Polynomial Size Neural Networks

被引：7

作者：

Obradovic, Zoran ^{[1
]}

Yan, Peiyuan ^{[2
]}

机构：

[1] Penn State Univ, Dept Comp Sci, University Pk, PA 16802 USA

[2] Lycoming Coll, Math Dept, Williamsport, PA 17701 USA

来源：

NEURAL COMPUTATION | 1990年 / 2卷 / 04期

关键词：

D O I：

10.1162/neco.1990.2.4.402

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

For polynomially bounded weights and polylogarithmic precision, analog neural networks of polynomial size and depth 3 are strictly more powerful than those of polynomial size and depth 2.

引用

页码：402 / 404

页数：3

共 5 条

[1]

Hajnal A., 1987, 28th Annual Symposium on Foundations of Computer Science (Cat. No.87CH2471-1), P99, DOI 10.1109/SFCS.1987.59

[2] MULTILAYER FEEDFORWARD NETWORKS ARE UNIVERSAL APPROXIMATORS [J].

HORNIK, K ;

STINCHCOMBE, M ;

WHITE, H .

NEURAL NETWORKS, 1989, 2 (05) :359-366

[3]

Obradovic Z., 1990, Machine Learning: Proceedings of the Seventh International Conference (1990), P392

[4]

Obradovic Z., 1990, CS9028 PENNS STAT U

[5]

Obradovic Z., 1990, ADV NEURAL INFORM PR, P702

← 1 →