I obtained my PhD in 2023 from EPFL under the supervision of Prof. Martin Jaggi in the laboratory of Machine Learning and Optimization. My research focuses on understanding attention-based models and applying transformers to different data modalities. I also had the chance to collaborate closely with Andreas Loukas on graph related topics. I received my Bachelor and Master degrees in Communication Systems from EPFL and previously interned with Amazon Berlin and Google Brain Berlin.
Publications
Attention is Not All You Need: Pure Attention Loses Rank Doubly Exponentially with Depth
Yihe Dong, Jean-Baptiste Cordonnier, Andreas Loukas
ICML 2021 (long oral)
•
Paper
•
Code
•
BibTex
@article{DBLP:journals/corr/abs-2103-03404,
author = {Yihe Dong and
Jean{-}Baptiste Cordonnier and
Andreas Loukas},
title = {Attention is Not All You Need: Pure Attention Loses Rank Doubly Exponentially
with Depth},
journal = {CoRR},
volume = {abs/2103.03404},
year = {2021},
url = {https://arxiv.org/abs/2103.03404},
archivePrefix = {arXiv},
}
Differentiable Patch Selection for Image Recognition
Jean-Baptiste Cordonnier*, Aravindh Mahendran*, Alexey Dosovitskiy, Dirk Weissenborn, Jakob Uszkoreit,
Thomas Unterthiner
CVPR 2021
•
Paper
•
Code
•
Slides
•
BibTex
@misc{cordonnier2021differentiable,
title={Differentiable Patch Selection for Image Recognition},
author={Jean-Baptiste Cordonnier and Aravindh Mahendran and Alexey Dosovitskiy and Dirk Weissenborn and Jakob Uszkoreit and Thomas Unterthiner},
year={2021},
eprint={2104.03059},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
Group Equivariant Stand-Alone Self-Attention For Vision
David W. Romero*, Jean-Baptiste Cordonnier*
ICLR 2021
•
Paper
•
Code
•
BibTex
@inproceedings{
romero2021group,
title={Group Equivariant Stand-Alone Self-Attention For Vision},
author={David W. Romero and Jean-Baptiste Cordonnier},
booktitle={International Conference on Learning Representations},
year={2021},
url={https://openreview.net/forum?id=JkfYjnOEo6M}
}
On the Relationship between Self-Attention and Convolutional Layers
Jean-Baptiste Cordonnier, Andreas Loukas and Martin Jaggi
ICLR 2020
•
Paper
•
Code
•
Blog
•
Explore
•
Slides
•
BibTex
@inproceedings{
cordonnier2020on,
title={On the Relationship between Self-Attention and Convolutional Layers},
author={Jean-Baptiste Cordonnier and Andreas Loukas and Martin Jaggi},
booktitle={International Conference on Learning Representations},
year={2020},
url={https://openreview.net/forum?id=HJlnC1rKPB}
}
Extrapolating Paths with Graph Neural Networks
Jean-Baptiste Cordonnier and Andreas Loukas
IJCAI 2019
•
Paper
•
Code
•
BibTex
@inproceedings{DBLP:conf/ijcai/CordonnierL19,
author = {Jean{-}Baptiste Cordonnier and
Andreas Loukas},
title = {Extrapolating Paths with Graph Neural Networks},
booktitle = {Proceedings of the Twenty-Eighth International Joint Conference on
Artificial Intelligence, {IJCAI} 2019, Macao, China, August 10-16,
2019},
pages = {2187--2194},
year = {2019}
}
Sparsified SGD with Memory
Sebastian U. Stich, Jean-Baptiste Cordonnier and Martin Jaggi
NeurIPS 2018
•
Paper
•
Poster
•
Code
•
BibTex
@inproceedings{DBLP:conf/nips/StichCJ18,
author = {Sebastian U. Stich and
Jean{-}Baptiste Cordonnier and
Martin Jaggi},
title = {Sparsified {SGD} with Memory},
booktitle = {NeurIPS},
pages = {4452--4463},
year = {2018}
}
Convex Optimization using Sparsified SGD with Memory
Jean-Baptiste Cordonnier
Master thesis EPFL
•
Paper
•
BibTex
@article{Cordonnier:262760,
title = {Convex Optimization using Sparsified Stochastic Gradient Descent with Memory},
author = {Cordonnier, Jean-Baptiste},
pages = {49 pages},
year = {2018},
url = {http://infoscience.epfl.ch/record/262760},
}
Projects
Avalanche Accidents Visualization
Jean-Baptiste Cordonnier, Brune Bastide and Arnaud Lesimple
AMLD 2018
•
Website
Best project award in Applied Data Analysis course