Publication:
Self-organized variational autoencoders (self-vae) for learned image compression

dc.contributor.coauthorMalik, J.
dc.contributor.coauthorKıranyaz S.
dc.contributor.departmentDepartment of Electrical and Electronics Engineering
dc.contributor.departmentGraduate School of Sciences and Engineering
dc.contributor.kuauthorGüven, Hilal
dc.contributor.kuauthorKeleş, Onur
dc.contributor.kuauthorTekalp, Ahmet Murat
dc.contributor.kuauthorYılmaz, Mustafa Akın
dc.contributor.schoolcollegeinstituteCollege of Engineering
dc.contributor.schoolcollegeinstituteGRADUATE SCHOOL OF SCIENCES AND ENGINEERING
dc.date.accessioned2024-11-09T13:50:24Z
dc.date.issued2021
dc.description.abstractIn end-to-end optimized learned image compression, it is standard practice to use a convolutional variational autoencoder with generalized divisive normalization (GDN) to transform images into a latent space. Recently, Operational Neural Networks (ONNs) that learn the best non-linearity from a set of alternatives, and their “self-organized” variants, Self-ONNs, that approximate any non-linearity via Taylor series have been proposed to address the limitations of convolutional layers and a fixed nonlinear activation. In this paper, we propose to replace the convolutional and GDN layers in the variational autoencoder with self-organized operational layers, and propose a novel self-organized variational autoencoder (Self-VAE) architecture that benefits from stronger non-linearity. The experimental results demonstrate that the proposed Self-VAE yields improvements in both rate-distortion performance and perceptual image quality.
dc.description.fulltextYES
dc.description.indexedbyWOS
dc.description.indexedbyScopus
dc.description.openaccessYES
dc.description.publisherscopeInternational
dc.description.sponsoredbyTubitakEuTÜBİTAK
dc.description.sponsorshipScientific and Technological Research Council of Turkey (TÜBİTAK)
dc.description.sponsorshipKoç Üniversitesi İş Bankası Yapay Zeka Uygulama ve Araştırma Merkezi (KUIS AI)/ Koç University İş Bank Artificial Intelligence Center (KUIS AI)
dc.description.sponsorshipTurkish Academy of Sciences (TÜBA)
dc.description.versionPublisher version
dc.identifier.doi10.1109/ICIP42928.2021.9506041
dc.identifier.embargoNO
dc.identifier.filenameinventorynoIR03568
dc.identifier.isbn9.78167E+12
dc.identifier.issn15224880
dc.identifier.quartileN/A
dc.identifier.scopus2-s2.0-85125590801
dc.identifier.urihttps://hdl.handle.net/20.500.14288/3908
dc.identifier.wos819455103170
dc.keywordsEnd-to-end learned image compression
dc.keywordsPerceptual quality metrics
dc.keywordsRate-distortion performance
dc.keywordsSelf-organized operational layer
dc.keywordsVariational autoencoder
dc.language.isoeng
dc.publisherInstitute of Electrical and Electronics Engineers (IEEE)
dc.relation.grantno120C156, 217E033
dc.relation.ispartofProceedings - International Conference on Image Processing, ICIP
dc.relation.urihttp://cdm21054.contentdm.oclc.org/cdm/ref/collection/IR/id/10434
dc.subjectCompression
dc.subjectJPEG
dc.subjectDeep learning
dc.titleSelf-organized variational autoencoders (self-vae) for learned image compression
dc.typeConference Proceeding
dspace.entity.typePublication
local.contributor.kuauthorTekalp, Ahmet Murat
local.contributor.kuauthorKeleş, Onur
local.contributor.kuauthorYılmaz, Mustafa Akın
local.contributor.kuauthorGüven, Hilal
local.publication.orgunit1College of Engineering
local.publication.orgunit1GRADUATE SCHOOL OF SCIENCES AND ENGINEERING
local.publication.orgunit2Department of Electrical and Electronics Engineering
local.publication.orgunit2Graduate School of Sciences and Engineering
relation.isOrgUnitOfPublication21598063-a7c5-420d-91ba-0cc9b2db0ea0
relation.isOrgUnitOfPublication3fc31c89-e803-4eb1-af6b-6258bc42c3d8
relation.isOrgUnitOfPublication.latestForDiscovery21598063-a7c5-420d-91ba-0cc9b2db0ea0
relation.isParentOrgUnitOfPublication8e756b23-2d4a-4ce8-b1b3-62c794a8c164
relation.isParentOrgUnitOfPublication434c9663-2b11-4e66-9399-c863e2ebae43
relation.isParentOrgUnitOfPublication.latestForDiscovery8e756b23-2d4a-4ce8-b1b3-62c794a8c164

Files

Original bundle

Now showing 1 - 1 of 1
Thumbnail Image
Name:
10434.pdf
Size:
5.1 MB
Format:
Adobe Portable Document Format