Completion of Missing Labels for Multi-Label Annotation by a Unified Graph Laplacian Regularization
IEICE Transactions on Information and Systems Volume E103.D Issue 10
Page 2154-2161
published_at 2020-10-01
アクセス数 : 152 件
ダウンロード数 : 40 件
今月のアクセス数 : 0 件
今月のダウンロード数 : 0 件
この文献の参照には次のURLをご利用ください : https://ir.lib.hiroshima-u.ac.jp/00051543
File |
IEICETIS_E103.D_2154.pdf
1.47 MB
種類 :
fulltext
|
Title ( eng ) |
Completion of Missing Labels for Multi-Label Annotation by a Unified Graph Laplacian Regularization
|
Creator |
MOJOO Jonathan
ZHAO Yu
KAVITHA Muthu Subash
|
Source Title |
IEICE Transactions on Information and Systems
|
Volume | E103.D |
Issue | 10 |
Start Page | 2154 |
End Page | 2161 |
Abstract |
The task of image annotation is becoming enormously important for efficient image retrieval from the web and other large databases. However, huge semantic information and complex dependency of labels on an image make the task challenging. Hence determining the semantic similarity between multiple labels on an image is useful to understand any incomplete label assignment for image retrieval. This work proposes a novel method to solve the problem of multi-label image annotation by unifying two different types of Laplacian regularization terms in deep convolutional neural network (CNN) for robust annotation performance. The unified Laplacian regularization model is implemented to address the missing labels efficiently by generating the contextual similarity between labels both internally and externally through their semantic similarities, which is the main contribution of this study. Specifically, we generate similarity matrices between labels internally by using Hayashi's quantification method-type III and externally by using the word2vec method. The generated similarity matrices from the two different methods are then combined as a Laplacian regularization term, which is used as the new objective function of the deep CNN. The Regularization term implemented in this study is able to address the multi-label annotation problem, enabling a more effectively trained neural network. Experimental results on public benchmark datasets reveal that the proposed unified regularization model with deep CNN produces significantly better results than the baseline CNN without regularization and other state-of-the-art methods for predicting missing labels.
|
Keywords |
multi-label image annotation
regularization
missing labels
|
Descriptions |
This work was partly supported by JSPS KAKENHI Grant Number 16K00239.
|
Language |
eng
|
Resource Type | journal article |
Publisher |
The Institute of Electronics, Information and Communication Engineers
電子情報通信学会
|
Date of Issued | 2020-10-01 |
Rights |
Copyright © 2020 The Institute of Electronics, Information and Communication Engineers
|
Publish Type | Version of Record |
Access Rights | open access |
Source Identifier |
[ISSN] 0916-8532
[ISSN] 1745-1361
[DOI] 10.1587/transinf.2019EDP7318
[DOI] https://doi.org/10.1587/transinf.2019EDP7318
|