Limitations of Shallow Networks Representing Finite Mappings

Result code in IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F67985807%3A_____%2F19%3A00485613" target="_blank" >RIV/67985807:_____/19:00485613 - isvavai.cz</a>

Result on the web

<a href="http://dx.doi.org/10.1007/s00521-018-3680-1" target="_blank" >http://dx.doi.org/10.1007/s00521-018-3680-1</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1007/s00521-018-3680-1" target="_blank" >10.1007/s00521-018-3680-1</a>

Result language

angličtina

Original language name

Limitations of Shallow Networks Representing Finite Mappings

Original language description

Limitations of capabilities of shallow networks to efficiently compute real-valued functions on finite domains are investigated. Efficiency is studied in terms of network sparsity and its approximate measures. It is shown that when a dictionary of computational units is not sufficiently large, computation of almost any uniformly randomly chosen function either represents a well-conditioned task performed by a large network or an ill-conditioned task performed by a network of a moderate size. The probabilistic results are complemented by a concrete example of a class of functions which cannot be efficiently computed by shallow perceptron networks. The class is constructed using pseudo-noise sequences which have many features of random sequences but can be generated using special polynomials. Connections to the No Free Lunch Theorem and the central paradox of coding theory are discussed.

Czech name

—

Czech description

—

Type

J_imp - Article in a specialist periodical, which is included in the Web of Science database

CEP classification

—

OECD FORD branch

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Project

Result was created during the realization of more than one project. More information in the Projects tab.

Continuities

I - Institucionalni podpora na dlouhodoby koncepcni rozvoj vyzkumne organizace

Publication year

2019

Confidentiality

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Name of the periodical

Neural Computing & Applications

ISSN

0941-0643

e-ISSN

—

Volume of the periodical

31

Issue of the periodical within the volume

6

Country of publishing house

US - UNITED STATES

Number of pages

10

Pages from-to

1783-1792

UT code for WoS article

000470746700008

EID of the result in the Scopus database

2-s2.0-85052492938