Nikos Komodakis

Timeline

1. 2006

   PhD at the University of Crete

   He completes his PhD in 2006 at the Computer Science Department of the University of Crete. He also works as a postdoctoral researcher at École Centrale Paris.

2. 2007–2011

   Dual decomposition for MRFs

   He researches optimization algorithms for Markov random fields and their uses in computer vision, medical imaging and computer graphics. In 2007, he gives a tutorial on discrete optimization methods in computer vision at the International Conference on Computer Vision in Rio de Janeiro. In 2011, he publishes "MRF Energy Minimization and Beyond via Dual Decomposition" with Nikos Paragios and Georgios Tziritas in IEEE Transactions on Pattern Analysis and Machine Intelligence.

3. By 2013

   Associate professor at École des Ponts

   He is an associate professor in the Department of Computer Science and Applied Math at École des Ponts ParisTech. In 2012, he gives a keynote at the ECCV 2012 workshop on higher-order models and global constraints in Firenze, and in 2013 he gives a tutorial on discrete optimization in computer vision at ICIAP in Naples.

4. 2015–2017

   Wide residual networks

   In 2015, he and Sergey Zagoruyko show how a convolutional network can learn a similarity function for comparing image patches directly from image data. In 2016, they propose wide residual networks, where a 16-layer-deep network outperforms thousand-layer-deep residual networks and sets new state-of-the-art results on CIFAR, SVHN and COCO. Their attention-transfer method, published at ICLR 2017, trains a student CNN to mimic the attention maps of a teacher network.

5. 2018

   Learning from image rotations

   With Spyros Gidaris and Praveer Singh, he trains ConvNets to recognize the 2D rotation applied to an input image, a task that supplies a supervisory signal for learning image features without labels; the work is accepted at ICLR 2018. With Gidaris, he also builds a few-shot learning system that learns new categories from a few examples without forgetting its base categories, accepted at CVPR 2018.

6. 2025

   Joint image-feature diffusion

   In 2025, he coauthors "Boosting Generative Image Modeling via Joint Image-Feature Synthesis" with Theodoros Kouzelis, Efstathios Karypidis, Ioannis Kakogeorgiou and Spyros Gidaris. A diffusion model jointly models image latents from a variational autoencoder and semantic features from an encoder such as DINO, and the paper is a NeurIPS 2025 spotlight.

7. As of October 2026

   Generative AI group at Archimedes

   He is an associate professor at the Computer Science Department of the University of Crete, where he moves after his associate professorship at École des Ponts ParisTech. He is also a leading researcher and head of the Generative AI group at the Archimedes Research Unit of the Athena Research Center, an associated faculty member at FORTH's Institute of Applied and Computational Mathematics, and an editorial board member of the International Journal of Computer Vision, Computer Vision and Image Understanding and Computational Intelligence.