Reference TypeConference Proceedings
Citation KeyOliveiraPenaBert:2020:CoGrSe
Author1 Oliveira, Willian Dihanster G. de
2 Penatti, Otávio A. B.
3 Berton, Lilian
Affiliation1 Federal University of Sao Paulo
2 Samsung R&D Institute
3 Federal University of Sao Paulo
TitleA comparison of graph-based semi-supervised learning for data augmentation
Conference NameConference on Graphics, Patterns and Images, 33 (SIBGRAPI)
EditorMusse, Soraia Raupp
Cesar Junior, Roberto Marcondes
Pelechano, Nuria
Wang, Zhangyang (Atlas)
Book TitleProceedings
DateNov. 7-10, 2020
Publisher CityLos Alamitos
PublisherIEEE Computer Society
Conference LocationVirtual
KeywordsImage classification, data augmentation, image transformation, GANs, semi-supervised learning, machine learning.
AbstractIn supervised learning, the algorithm accuracy usually improves with the size of the labeled dataset used for training the classifier. However, in many real-life scenarios, obtaining enough labeled data is costly or even not possible. In many circumstances, Data Augmentation (DA) techniques are usually employed, generating more labeled data for training machine learning algorithms. The common DA techniques are applied to already labeled data, generating simple variations of this data. For example, for image classification, image samples are rotated, cropped, flipped or other operators to generate variations of input image samples, and keeping their original labels. Other options are using Neural Networks algorithms that create new synthetic data or to employ Semi-supervised Learning (SSL) that label existing unlabeled data. In this paper, we perform a comparison among graph-based semi-supervised learning (GSSL) algorithms to augment the labeled dataset. The main advantage of using GSSL is that we can increase the training set by adding non-annotated images to the training set, therefore, we can benefit from the huge amount of unlabeled data available. Experiments are performed on five datasets for recognition of handwritten digits and letters (MNIST and EMINIST), animals (Dogs vs Cats), clothes (MNIST-Fashion) and remote sensing images (Brazilian Coffee Scenes), in which we compare different possibilities for DA, including the GSSL, Generative Adversarial Networks (GANs) and traditional Image Transformations (IT) applied on input labeled data. We also evaluated the impact of such techniques on different convolutional neural networks (CNN). Results indicate that, although all DA techniques performed well, GSSL was more robust to different image properties, presenting less accuracy variation across datasets.
Tertiary TypeFull Paper
Size333 KiB
Number of Files1
Target Filesibgrapi2020_ID30.pdf
Last Update2020:
Metadata Last Update2020: {D 2020}
Document Stagecompleted
Is the master or a copy?is the master
e-Mail (login)
Content TypeExternal Contribution
Document Stagenot transferred
source Directory Content
Sibgrapi2020_ID30.pdf 24/09/2020 16:34 561.9 KiB 
agreement Directory Content
agreement.html 24/09/2020 16:34 1.2 KiB 
Empty Fieldsaccessionnumber archivingpolicy archivist area callnumber copyholder copyright creatorhistory descriptionlevel dissemination doi edition electronicmailaddress group holdercode isbn issn label lineage mark nextedition nexthigherunit notes numberofvolumes orcid organization pages parameterlist parentrepositories previousedition previouslowerunit progress project readergroup readpermission resumeid rightsholder secondarydate secondarykey secondarymark secondarytype serieseditor session shorttitle sponsor subject tertiarymark type url versiontype volume
Access Date2020, Oct. 24