Deep Dropout Artificial Neural Networks for Recognising Digits and Characters in Natural Images

Erik Barrow, Chrisina Jayne, Mark Eastwood

Research output: Chapter in Book/Report/Conference proceedingChapter

1 Citation (Scopus)

Abstract

Recognising images using computers is a traditionally hard problem in computing, and one that becomes particularly difficult when these images are from the real world due to the large variations in them. This paper investigates the problem of recognising digits and characters in natural images using a deep neural network approach. The experiments explore the utilisation of a recently introduced dropout method which reduces overfitting. A number of different configuration networks are trained. It is found that the majority of networks give better accuracy when trained using the dropout method. This indicates that dropout is an effective method to improve training of deep neural networks on the application of recognising natural images of digits and characters.
Original languageEnglish
Title of host publicationNeural Information Processing
EditorsSabri Arik, Tingwen Huang, Weng Kin Lai, Qingshan Liu
Place of PublicationSwitzerland
PublisherSpringer Verlag
Pages29-37
Volume9492
ISBN (Electronic)978-3-319-26561-2
ISBN (Print)978-3-319-26560-5
DOIs
Publication statusPublished - 18 Nov 2015
Event22nd International Conference on Neural Information Processing (ICONIP2015) - Istanbul, Turkey
Duration: 9 Nov 201512 Nov 2015

Publication series

NameLecture Notes in Computer Science
PublisherSpringer
Volume9492
ISSN (Print)0302-9743

Conference

Conference22nd International Conference on Neural Information Processing (ICONIP2015)
CountryTurkey
CityIstanbul
Period9/11/1512/11/15

Fingerprint

Neural networks
Experiments
Deep neural networks

Bibliographical note

There is no full text available.

Keywords

  • Character recognition
  • Natural images
  • Artificial neural network
  • Deep learning
  • Dropout network

Cite this

Barrow, E., Jayne, C., & Eastwood, M. (2015). Deep Dropout Artificial Neural Networks for Recognising Digits and Characters in Natural Images. In S. Arik, T. Huang, W. K. Lai, & Q. Liu (Eds.), Neural Information Processing (Vol. 9492, pp. 29-37). (Lecture Notes in Computer Science; Vol. 9492). Switzerland: Springer Verlag. https://doi.org/10.1007/978-3-319-26561-2_4

Deep Dropout Artificial Neural Networks for Recognising Digits and Characters in Natural Images. / Barrow, Erik; Jayne, Chrisina; Eastwood, Mark.

Neural Information Processing. ed. / Sabri Arik; Tingwen Huang; Weng Kin Lai; Qingshan Liu. Vol. 9492 Switzerland : Springer Verlag, 2015. p. 29-37 (Lecture Notes in Computer Science; Vol. 9492).

Research output: Chapter in Book/Report/Conference proceedingChapter

Barrow, E, Jayne, C & Eastwood, M 2015, Deep Dropout Artificial Neural Networks for Recognising Digits and Characters in Natural Images. in S Arik, T Huang, WK Lai & Q Liu (eds), Neural Information Processing. vol. 9492, Lecture Notes in Computer Science, vol. 9492, Springer Verlag, Switzerland, pp. 29-37, 22nd International Conference on Neural Information Processing (ICONIP2015), Istanbul, Turkey, 9/11/15. https://doi.org/10.1007/978-3-319-26561-2_4
Barrow E, Jayne C, Eastwood M. Deep Dropout Artificial Neural Networks for Recognising Digits and Characters in Natural Images. In Arik S, Huang T, Lai WK, Liu Q, editors, Neural Information Processing. Vol. 9492. Switzerland: Springer Verlag. 2015. p. 29-37. (Lecture Notes in Computer Science). https://doi.org/10.1007/978-3-319-26561-2_4
Barrow, Erik ; Jayne, Chrisina ; Eastwood, Mark. / Deep Dropout Artificial Neural Networks for Recognising Digits and Characters in Natural Images. Neural Information Processing. editor / Sabri Arik ; Tingwen Huang ; Weng Kin Lai ; Qingshan Liu. Vol. 9492 Switzerland : Springer Verlag, 2015. pp. 29-37 (Lecture Notes in Computer Science).
@inbook{903cb2b1c8674999bb4751d0c13b0160,
title = "Deep Dropout Artificial Neural Networks for Recognising Digits and Characters in Natural Images",
abstract = "Recognising images using computers is a traditionally hard problem in computing, and one that becomes particularly difficult when these images are from the real world due to the large variations in them. This paper investigates the problem of recognising digits and characters in natural images using a deep neural network approach. The experiments explore the utilisation of a recently introduced dropout method which reduces overfitting. A number of different configuration networks are trained. It is found that the majority of networks give better accuracy when trained using the dropout method. This indicates that dropout is an effective method to improve training of deep neural networks on the application of recognising natural images of digits and characters.",
keywords = "Character recognition, Natural images, Artificial neural network, Deep learning, Dropout network",
author = "Erik Barrow and Chrisina Jayne and Mark Eastwood",
note = "There is no full text available.",
year = "2015",
month = "11",
day = "18",
doi = "10.1007/978-3-319-26561-2_4",
language = "English",
isbn = "978-3-319-26560-5",
volume = "9492",
series = "Lecture Notes in Computer Science",
publisher = "Springer Verlag",
pages = "29--37",
editor = "Sabri Arik and Tingwen Huang and Lai, {Weng Kin} and Qingshan Liu",
booktitle = "Neural Information Processing",
address = "Austria",

}

TY - CHAP

T1 - Deep Dropout Artificial Neural Networks for Recognising Digits and Characters in Natural Images

AU - Barrow, Erik

AU - Jayne, Chrisina

AU - Eastwood, Mark

N1 - There is no full text available.

PY - 2015/11/18

Y1 - 2015/11/18

N2 - Recognising images using computers is a traditionally hard problem in computing, and one that becomes particularly difficult when these images are from the real world due to the large variations in them. This paper investigates the problem of recognising digits and characters in natural images using a deep neural network approach. The experiments explore the utilisation of a recently introduced dropout method which reduces overfitting. A number of different configuration networks are trained. It is found that the majority of networks give better accuracy when trained using the dropout method. This indicates that dropout is an effective method to improve training of deep neural networks on the application of recognising natural images of digits and characters.

AB - Recognising images using computers is a traditionally hard problem in computing, and one that becomes particularly difficult when these images are from the real world due to the large variations in them. This paper investigates the problem of recognising digits and characters in natural images using a deep neural network approach. The experiments explore the utilisation of a recently introduced dropout method which reduces overfitting. A number of different configuration networks are trained. It is found that the majority of networks give better accuracy when trained using the dropout method. This indicates that dropout is an effective method to improve training of deep neural networks on the application of recognising natural images of digits and characters.

KW - Character recognition

KW - Natural images

KW - Artificial neural network

KW - Deep learning

KW - Dropout network

U2 - 10.1007/978-3-319-26561-2_4

DO - 10.1007/978-3-319-26561-2_4

M3 - Chapter

SN - 978-3-319-26560-5

VL - 9492

T3 - Lecture Notes in Computer Science

SP - 29

EP - 37

BT - Neural Information Processing

A2 - Arik, Sabri

A2 - Huang, Tingwen

A2 - Lai, Weng Kin

A2 - Liu, Qingshan

PB - Springer Verlag

CY - Switzerland

ER -