Valentinos Pariza

I work with Yuki M. Asano at the Fundamental AI Lab.

I study how models understand images and how that knowledge can help them generate and edit images. My earlier work focused on self-supervised vision encoders and the spatial information in their features. These days, I work on diffusion models for perception, editing, and generation, with growing interests in multimodal learning and video. Across these projects, I want to understand what visual information carries over between tasks and how to train and adapt models efficiently.

Background

I completed my MSc in Artificial Intelligence at the University of Amsterdam, where I graduated cum laude and took part in the ELLIS MSc Honours Programme. My thesis was a collaboration between UvA, TNO, and ISTA, supervised by Yuki M. Asano, Gertjan Burghouts, and Francesco Locatello. Before that, I studied Computer Science at the University of Cyprus.

NeCo grew out of my MSc thesis. During that work, I also built Open Hummingbird Evaluation, an open-source implementation of Balažević et al.’s dense nearest-neighbor evaluation.

I’ve also worked at Amazon, South Pole, and CYENS, and co-founded Fooderloo to help reduce food waste in Cyprus. My CV has more about these roles.

Outside research, I enjoy music, cycling, other sports, and travelling.

Recent news

All news