Rates of convergence for fast learning RBF neural nets

Diego Attilio Mancuso

Rates of convergence for fast learning RBF neural nets

Risultato della ricerca: Contributo in libro › Contributo a convegno

Abstract

[Ita:]Le reti neurali sono modelli di regressione non parametrica interessanti per la loro efficacia in presenza di un elevato numero di variabili esplicative. Tra i vari schemi di rete neurale quello di tipo RBF ad apprendimento rapido si caratterizza per un processo di stima estremamente semplice. In questo lavoro viene esplicitata la velocità di convergenza delle reti RBF ad apprendimento rapido mettendo in evidenza come, sotto tenui ipotesi di regolarità, questa non dipenda dal numero di regressori utilizzati.

Lingua originale	English
Titolo della pubblicazione ospite	XLII Riunione Scientifica Bari, 9-11 giugno 2004
Pagine	59-62
Numero di pagine	4
Stato di pubblicazione	Pubblicato - 2004
Evento	XLII Riunione scientifica SIS - Bari Durata: 9 giu 2004 → 11 giu 2004

Convegno

Convegno	XLII Riunione scientifica SIS
Città	Bari
Periodo	9/6/04 → 11/6/04

Keywords

Radial Basis Functions
neural networks
nonparametric regression

Altri file e collegamenti

Link a IRIS PubliCatt

Cita questo

@inproceedings{03942907e13b4bb6b643e9dae81905c2,

title = "Rates of convergence for fast learning RBF neural nets",

abstract = "[Autom. eng. transl.] Neural networks are non-parametric regression models of interest for their effectiveness in the presence of a large number of explanatory variables. Among the various neural network schemes, the RBF type with rapid learning is characterized by an extremely simple estimation process. In this work the convergence speed of rapid learning RBF networks is made explicit by highlighting how, under tenuous regularity hypotheses, this does not depend on the number of regressors used.",

keywords = "Radial Basis Functions, neural networks, nonparametric regression, Radial Basis Functions, neural networks, nonparametric regression",

author = "Mancuso, {Diego Attilio}",

year = "2004",

language = "English",

isbn = "88-7178-034-5",

pages = "59--62",

booktitle = "XLII Riunione Scientifica Bari, 9-11 giugno 2004",

note = "XLII Riunione scientifica SIS ; Conference date: 09-06-2004 Through 11-06-2004",

}

TY - GEN

T1 - Rates of convergence for fast learning RBF neural nets

AU - Mancuso, Diego Attilio

PY - 2004

Y1 - 2004

N2 - [Autom. eng. transl.] Neural networks are non-parametric regression models of interest for their effectiveness in the presence of a large number of explanatory variables. Among the various neural network schemes, the RBF type with rapid learning is characterized by an extremely simple estimation process. In this work the convergence speed of rapid learning RBF networks is made explicit by highlighting how, under tenuous regularity hypotheses, this does not depend on the number of regressors used.

AB - [Autom. eng. transl.] Neural networks are non-parametric regression models of interest for their effectiveness in the presence of a large number of explanatory variables. Among the various neural network schemes, the RBF type with rapid learning is characterized by an extremely simple estimation process. In this work the convergence speed of rapid learning RBF networks is made explicit by highlighting how, under tenuous regularity hypotheses, this does not depend on the number of regressors used.

KW - Radial Basis Functions

KW - neural networks

KW - nonparametric regression

KW - Radial Basis Functions

KW - neural networks

KW - nonparametric regression

UR - http://hdl.handle.net/10807/21465

M3 - Conference contribution

SN - 88-7178-034-5

SP - 59

EP - 62

BT - XLII Riunione Scientifica Bari, 9-11 giugno 2004

T2 - XLII Riunione scientifica SIS

Y2 - 9 June 2004 through 11 June 2004

ER -

Rates of convergence for fast learning RBF neural nets

Abstract

Convegno

Keywords

Altri file e collegamenti

Fingerprint

Cita questo