Mert Bulent Sariyildiz

I am a PhD student at NAVER LABS Europe and Inria Grenoble (the THOTH Team) in France, working with my amazing supervisors Yannis Kalantidis, Diane Larlus and Karteek Alahari. My PhD focuses on learning general-purpose visual representations from images.

I received my M.Sc. degree from the Computer Engineering Department at Bilkent University in Turkey. During my master, I worked with Gokberk Cinbis on learning data-efficient visual embedding models. I received my B.Sc. from the Electrical and Electronics Engineering Department at Anadolu (now Eskisehir Technical) University in Turkey.

Email  /  CV  /  Google Scholar  /  Twitter  /  Github

profile photo


I'm interested in computer vision, more specifically, self-supervised learning (SSL) and vision & language (VL). In SSL, the goal is to learn general-purpose visual representations (prior knowledge about visual world) without relying on human-defined semantic annotations. I try to understand what makes general-purpose representations and how to learn them with only unlabeled data. VL, on the other hand, tries to align or complement visual and accompanying textual data representations. I find it exciting to study how language can provide prior knowledge to learn visual representations that can extrapolate to unseen visual concepts.

in1k vs transfer plane Improving the Generalization of Supervised Models
Mert Bulent Sariyildiz, Yannis Kalantidis, Karteek Alahari, Diane Larlus

We revisit supervised learning on ImageNet-1K and propose a training setup which improves transfer learning performance of supervised models.
project website

hypothetical-cog-levels ImageNet-CoG | Concept Generalization in Visual Representation Learning
Mert Bulent Sariyildiz, Yannis Kalantidis, Diane Larlus, Karteek Alahari
ICCV 2021

We propose a benchmark tailored for measuring concept generalization capabilities of models.
project website, code, poster, presentation (PDF), presentation (PPT), video

mochi-mixing-negatives MoCHi | Hard Negative Mixing for Contrastive Learning
Yannis Kalantidis, Mert Bulent Sariyildiz, Noe Pion, Philippe Weinzaepfel, Diane Larlus
NeurIPS 2020

For contrastive learning, sampling more or harder negatives often improve performance. We propose two ways to synthesize more negatives using the MoCo framework.
project website

icmlm-masked-token-attention ICMLM | Learning Visual Representations with Caption Annotations
Mert Bulent Sariyildiz, Julien Perez, Diane Larlus
ECCV 2020

Images often come with accompanying text describing the scene in images. We propose a method to learn visual representations using (image, caption) pairs.
project website, demo

keyp-compute-chain Key protected classification for collaborative learning
Mert Bulent Sariyildiz, Ramazan Gokberk Cinbis, Erman Ayday
Pattern Recognition, Vol. 104, August 2020

Vanilla collaborative learning frameworks are vulnerable to an active adversary that runs a generative adversarial network attack. We propose a classification model that is resilient against such attacks by design.
code repo

gmn-front-figure GMN | Gradient Matching Generative Networks for Zero-Shot Learning
Mert Bulent Sariyildiz, Ramazan Gokberk Cinbis
CVPR 2019, oral presentation

Zero-shot learning models may suffer from the domain-shift due to the difference between data distributions of seen and unseen concepts. We propose a generative model to synthesize samples for unseen concepts given their visual attributes and use these samples for training a classifier for both seen and unseen concepts.
code repo

Community Service

Huge thanks to Jon Barron, who provides the template of this website.