Minimizing the Quadratic Training Error of a Sigmoid Neuron is Hard.

Result code in IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F67985807%3A_____%2F01%3A06010079" target="_blank" >RIV/67985807:_____/01:06010079 - isvavai.cz</a>

Result on the web

—

DOI - Digital Object Identifier

—

Result language

angličtina

Original language name

Minimizing the Quadratic Training Error of a Sigmoid Neuron is Hard.

Original language description

We first present a brief survey of hardness results for training feedforward neural networks. These results are then completed by the proof that the simplest architecture containing only a single neuron that applies the standard (logistic) activation function to the weighted sum of n inputs is hard to train. In particular,the problem of finding the weights of such a unit that minimize the relative quadratic training error within 1 or its average (over a training set) within 13/(31n) of its infimum proves...

Czech name

—

Czech description

—

Type

D - Article in proceedings

CEP classification

BA - General mathematics

OECD FORD branch

—

Project

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

Continuities

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Publication year

2001

Confidentiality

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

Article name in the collection

Algorithmic Learning Theory.

ISBN

3-540-42875-5

ISSN

—

e-ISSN

—

Number of pages

14

Pages from-to

92-105

Publisher name

Springer

Place of publication

Berlin

Event location

Washington [US]

Event date

Nov 25, 2001

Type of event by nationality

WRD - Celosvětová akce

UT code for WoS article

—