Visualization of neural-network gaps based on error analysis

Mehmed M. Kantardzic; Alaaeldin A. Aly; Adel S. Elmaghraby

doi:10.1109/72.750572

Visualization of neural-network gaps based on error analysis

Mehmed M. Kantardzic, Alaaeldin A. Aly, Adel S. Elmaghraby

Research output: Contribution to journal › Article › peer-review

6 Citations (Scopus)

Abstract

Presented in this paper is a new methodology for detection of neural-network gaps (NNG's) based on error analysis and the visualization that is applicable for n-dimensional I/O domain. The generalization problem in artificial neural networks (ANN) training is analyzed and the concept of NNG's is introduced. The NNG's are highly undesirable in ANN generalization and methods for detecting, analyzing, and eliminating them are necessary. Previous methods for NNG detection, based on two-dimensional (2-D) and three-dimensional (3-D) visualization, were not applicable for ANN's with more than three inputs. Experiments demonstrate advantages of this new methodology, which allows better understanding of the NNG phenomena using a quantitative approach.

Original language	English
Pages (from-to)	419-426
Number of pages	8
Journal	IEEE Transactions on Neural Networks
Volume	10
Issue number	2
DOIs	https://doi.org/10.1109/72.750572
Publication status	Published - Dec 1 1999

Keywords

Error analysis
Neural-network gaps
Neural-network generalization
Visualization

ASJC Scopus subject areas

Software
Computer Science Applications
Computer Networks and Communications
Artificial Intelligence

Access to Document

10.1109/72.750572

Cite this

@article{c4274f8908d54e0b953fa8c8a2699a50,

title = "Visualization of neural-network gaps based on error analysis",

abstract = "Presented in this paper is a new methodology for detection of neural-network gaps (NNG's) based on error analysis and the visualization that is applicable for n-dimensional I/O domain. The generalization problem in artificial neural networks (ANN) training is analyzed and the concept of NNG's is introduced. The NNG's are highly undesirable in ANN generalization and methods for detecting, analyzing, and eliminating them are necessary. Previous methods for NNG detection, based on two-dimensional (2-D) and three-dimensional (3-D) visualization, were not applicable for ANN's with more than three inputs. Experiments demonstrate advantages of this new methodology, which allows better understanding of the NNG phenomena using a quantitative approach.",

keywords = "Error analysis, Neural-network gaps, Neural-network generalization, Visualization",

author = "Kantardzic, {Mehmed M.} and Aly, {Alaaeldin A.} and Elmaghraby, {Adel S.}",

year = "1999",

month = dec,

day = "1",

doi = "10.1109/72.750572",

language = "English",

volume = "10",

pages = "419--426",

journal = "IEEE Transactions on Neural Networks",

issn = "1045-9227",

publisher = "IEEE Computational Intelligence Society",

number = "2",

}

TY - JOUR

T1 - Visualization of neural-network gaps based on error analysis

AU - Kantardzic, Mehmed M.

AU - Aly, Alaaeldin A.

AU - Elmaghraby, Adel S.

PY - 1999/12/1

Y1 - 1999/12/1

N2 - Presented in this paper is a new methodology for detection of neural-network gaps (NNG's) based on error analysis and the visualization that is applicable for n-dimensional I/O domain. The generalization problem in artificial neural networks (ANN) training is analyzed and the concept of NNG's is introduced. The NNG's are highly undesirable in ANN generalization and methods for detecting, analyzing, and eliminating them are necessary. Previous methods for NNG detection, based on two-dimensional (2-D) and three-dimensional (3-D) visualization, were not applicable for ANN's with more than three inputs. Experiments demonstrate advantages of this new methodology, which allows better understanding of the NNG phenomena using a quantitative approach.

AB - Presented in this paper is a new methodology for detection of neural-network gaps (NNG's) based on error analysis and the visualization that is applicable for n-dimensional I/O domain. The generalization problem in artificial neural networks (ANN) training is analyzed and the concept of NNG's is introduced. The NNG's are highly undesirable in ANN generalization and methods for detecting, analyzing, and eliminating them are necessary. Previous methods for NNG detection, based on two-dimensional (2-D) and three-dimensional (3-D) visualization, were not applicable for ANN's with more than three inputs. Experiments demonstrate advantages of this new methodology, which allows better understanding of the NNG phenomena using a quantitative approach.

KW - Error analysis

KW - Neural-network gaps

KW - Neural-network generalization

KW - Visualization

UR - http://www.scopus.com/inward/record.url?scp=0033097131&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0033097131&partnerID=8YFLogxK

U2 - 10.1109/72.750572

DO - 10.1109/72.750572

M3 - Article

C2 - 18252539

AN - SCOPUS:0033097131

SN - 1045-9227

VL - 10

SP - 419

EP - 426

JO - IEEE Transactions on Neural Networks

JF - IEEE Transactions on Neural Networks

IS - 2

ER -

Visualization of neural-network gaps based on error analysis

Abstract

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this