Sequential Convolutional Residual Network for Image Recognition

Wonjun HWANG

doi:10.1587/transinf.2017EDL8233

Sequential Convolutional Residual Network for Image Recognition

Wonjun HWANG

Full Text Views

0

Share
Cite this

Summary :

In this letter, we propose a sequential convolutional residual network, where we first analyze a tangled network architecture using simplified equations and determine the critical point to untangle the complex network architecture. Although the residual network shows good performance, the learning efficiency is not better than expected at deeper layers because the network is excessively intertwined. To solve this problem, we propose a network in which the information is transmitted sequentially. In this network architecture, the neighboring layer output adds the input of the current layer and iteratively passes its result to the next sequential layer. Thus, the proposed network can improve the learning efficiency and performance by successfully mitigating the complexity in deep networks. We show that the proposed network performs well on the Cifar-10 and Cifar-100 datasets. In particular, we prove that the proposed method is superior to the baseline method as the depth increases.

Publication: IEICE TRANSACTIONS on Information Vol.E101-D No.4 pp.1213-1216

Publication Date: 2018/04/01

Publicized: 2018/01/18

Online ISSN: 1745-1361

DOI: 10.1587/transinf.2017EDL8233

Type of Manuscript: LETTER

Category: Image Recognition, Computer Vision

Authors

Wonjun HWANG
Ajou University

Keyword

deep learning, convolutional neural network, residual network

Cite this

Copy

Wonjun HWANG, "Sequential Convolutional Residual Network for Image Recognition" in IEICE TRANSACTIONS on Information, vol. E101-D, no. 4, pp. 1213-1216, April 2018, doi: 10.1587/transinf.2017EDL8233.
Abstract: In this letter, we propose a sequential convolutional residual network, where we first analyze a tangled network architecture using simplified equations and determine the critical point to untangle the complex network architecture. Although the residual network shows good performance, the learning efficiency is not better than expected at deeper layers because the network is excessively intertwined. To solve this problem, we propose a network in which the information is transmitted sequentially. In this network architecture, the neighboring layer output adds the input of the current layer and iteratively passes its result to the next sequential layer. Thus, the proposed network can improve the learning efficiency and performance by successfully mitigating the complexity in deep networks. We show that the proposed network performs well on the Cifar-10 and Cifar-100 datasets. In particular, we prove that the proposed method is superior to the baseline method as the depth increases.
URL: https://globals.ieice.org/en_transactions/information/10.1587/transinf.2017EDL8233/_p

Copy

@ARTICLE{e101-d_4_1213,
author={Wonjun HWANG, },
journal={IEICE TRANSACTIONS on Information},
title={Sequential Convolutional Residual Network for Image Recognition},
year={2018},
volume={E101-D},
number={4},
pages={1213-1216},
abstract={In this letter, we propose a sequential convolutional residual network, where we first analyze a tangled network architecture using simplified equations and determine the critical point to untangle the complex network architecture. Although the residual network shows good performance, the learning efficiency is not better than expected at deeper layers because the network is excessively intertwined. To solve this problem, we propose a network in which the information is transmitted sequentially. In this network architecture, the neighboring layer output adds the input of the current layer and iteratively passes its result to the next sequential layer. Thus, the proposed network can improve the learning efficiency and performance by successfully mitigating the complexity in deep networks. We show that the proposed network performs well on the Cifar-10 and Cifar-100 datasets. In particular, we prove that the proposed method is superior to the baseline method as the depth increases.},
keywords={},
doi={10.1587/transinf.2017EDL8233},
ISSN={1745-1361},
month={April},}

Copy

TY - JOUR
TI - Sequential Convolutional Residual Network for Image Recognition
T2 - IEICE TRANSACTIONS on Information
SP - 1213
EP - 1216
AU - Wonjun HWANG
PY - 2018
DO - 10.1587/transinf.2017EDL8233
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E101-D
IS - 4
JA - IEICE TRANSACTIONS on Information
Y1 - April 2018
AB - In this letter, we propose a sequential convolutional residual network, where we first analyze a tangled network architecture using simplified equations and determine the critical point to untangle the complex network architecture. Although the residual network shows good performance, the learning efficiency is not better than expected at deeper layers because the network is excessively intertwined. To solve this problem, we propose a network in which the information is transmitted sequentially. In this network architecture, the neighboring layer output adds the input of the current layer and iteratively passes its result to the next sequential layer. Thus, the proposed network can improve the learning efficiency and performance by successfully mitigating the complexity in deep networks. We show that the proposed network performs well on the Cifar-10 and Cifar-100 datasets. In particular, we prove that the proposed method is superior to the baseline method as the depth increases.
ER -