Variational self-supervised learning
| dc.authorid | 0000-0003-1677-9496 | |
| dc.authorid | 0000-0001-7403-7592 | |
| dc.contributor.author | Yavuz, Mehmet Can | en_US |
| dc.contributor.author | Yanıkoğlu, Berrin | en_US |
| dc.date.accessioned | 2025-10-06T11:51:09Z | |
| dc.date.available | 2025-10-06T11:51:09Z | |
| dc.date.issued | 2025-04-06 | |
| dc.department | Işık Üniversitesi, Mühendislik ve Doğa Bilimleri Fakültesi, Bilgisayar Mühendisliği Bölümü | en_US |
| dc.department | Işık University, Faculty of Engineering and Natural Sciences, Department of Computer Engineering | en_US |
| dc.description.abstract | We present Variational Self-Supervised Learning (VSSL), a novel framework that combines variational inference with self-supervised learning to enable efficient, decoder-free representation learning. Unlike traditional VAEs that rely on input reconstruction via a decoder, VSSL symmetrically couples two encoders with Gaussian outputs. A momentum-updated teacher network defines a dynamic, data-dependent prior, while the student encoder produces an approximate posterior from augmented views. The reconstruction term in the ELBO is replaced with a cross-view denoising objective, preserving the analytical tractability of Gaussian KL divergence. We further introduce cosine-based formulations of KL and log-likelihood terms to enhance semantic alignment in high-dimensional latent spaces. Experiments on CIFAR-10, CIFAR-100, and ImageNet-100 show that VSSL achieves competitive or superior performance to leading self-supervised methods, including BYOL and MoCo V3. VSSL offers a scalable, probabilistically grounded approach to learning transferable representations without generative reconstruction, bridging the gap between variational modeling and modern self-supervised techniques. | en_US |
| dc.description.version | Preprint's Version | en_US |
| dc.identifier.citation | Yavuz, M. C. & Yanıkoğlu, B. (2025). Variational self-supervised learning. Arxiv, 1-6. doi:https://doi.org/10.48550/arXiv.2504.04318 | en_US |
| dc.identifier.endpage | 6 | |
| dc.identifier.startpage | 1 | |
| dc.identifier.uri | https://hdl.handle.net/11729/6745 | |
| dc.identifier.uri | https://doi.org/10.48550/arXiv.2504.04318 | |
| dc.identifier.wos | PPRN:122847704 | |
| dc.identifier.wosquality | N/A | |
| dc.indekslendigikaynak | Web of Science | en_US |
| dc.indekslendigikaynak | Preprint Citation Index | en_US |
| dc.institutionauthor | Yavuz, Mehmet Can | en_US |
| dc.institutionauthorid | 0000-0003-1677-9496 | |
| dc.language.iso | en | en_US |
| dc.publisher | Cornell Univ | en_US |
| dc.relation.ispartof | Arxiv | en_US |
| dc.relation.publicationcategory | Ön Baskı - Uluslararası - Kurum Öğretim Elemanı | en_US |
| dc.rights | info:eu-repo/semantics/openAccess | en_US |
| dc.subject | Self-supervised learning | en_US |
| dc.subject | Variational inference | en_US |
| dc.subject | Representation learning | en_US |
| dc.subject | Encoder-only models | en_US |
| dc.title | Variational self-supervised learning | en_US |
| dc.type | Preprint | en_US |
| dspace.entity.type | Publication | en_US |
