Pascal Mettes

Me and my team focus on deep learning in hyperbolic space. Within deep learning, we tend to question all aspects of training neural networks, from architectures and optimization to data and tricks. The most fundamental assumption, namely to operate in Euclidean space, is however rarely questioned. We believe that opening our scope beyond Euclid opens an entirely new worlds for deep learning. Specifically, we focus on hyperbolic deep learning. Learning in hyperbolic space enables us to learn hierarchical representations, with stronger robustness (with respect to OOD and adversarial samples), more compactness, and with the possibility of incorporating prior knowledge. We also consider it the native space for vision-language models. In our lab, we are therefore spearheading hyperbolic deep learning through algorithmic advances, open-source developments, and international workshops, tutorials, and talks.

This page will highlight some success stories of our journey, including the first software library for hyperbolic learning, the first international tutorials and survey on hyperbolic learning for computer vision, and hyperbolic learning for various domains, ranging from image segmentation and social navigation to hierarchical classification and vision-language alignment.

Recent news highlights

[07-2024] "Hyp2Nav: Hyperbolic Planning and Curiosity for Crowd Navigation" accepted to IROS [pdf].
[06-2024] "Intriguing Properties of Hyperbolic Vision-Language Models" accepted to TMLR [pdf].
[05-2024] "Hyperbolic Random Forests" accepted to TMLR [pdf].
[05-2024] Workshop on Hyperbolic and Hyperspherical Learning for Computer Vision at ECCV 2024!
[04-2024] I am a keynote at the 1st Workshop on Human-Centered Vision and Media Technologies.
[04-2024] I am a keynote at the CV papers challenge workshop hosted by AIST, Tokyo.
[03-2024] Our survey on Hyperbolic Deep Learning in Computer Vision has been accepted to IJCV.
[01-2024] I will be a speaker at the Oxford Machine Learning Summer School 2024.
[12-2023] Guido D'Amely and Alessandro Flaborea (University of Sapienza) will be visiting for 3 months.
[10-2023] "Poincaré ResNet" accepted to ICCV 2023 [pdf].
[08-2023] "HypLL: The Hyperbolic Learning Library" is accepted to ACM Multimedia 2023.
[06-2023] HypLL: The Hyperbolic Learning Library is now available! [github] [pdf]
[06-2023] We are organizing a new tutorial on hyperbolic learning at CVPR 2023.
[06-2023] I am giving a keynote at the CVPR 2023 anti-UAV workshop.
[05-2023] Our survey on Hyperbolic Deep Learning in Computer Vision is available online.
[02-2023] The recordings of the ECCV 2022 hyperbolic tutorial is now available online.

Organization and network

• 2024 - ECCV workshop on Hyperbolic and Hyperspherical Learning for Computer Vision
• 2024 - Area chair for CVPR 2024, ECCV 2024, NeurIPS 2024
• 2023 - CVPR tutorial on Hyperbolic Deep Learning for Computer Vision
• 2023 - Area chair for NeurIPS 2023
• 2023 - Area chair BMVC (since 2020)
• 2022 - ECCV tutorial on Hyperbolic Representation Learning for Computer Vision
• 2022 - Area chair for CVPR 2023
• 2022 - Video Understanding Symposium in Amsterdam
• 2022 - Panelist at the International Conference on Multimedia Retrieval
• 2022 - Netherlands Conference on Computer Vision
• 2021 - Member of ELLIS - European Laboratory for Learning and Intelligent Systems
• 2021 - ICCV workshop on Structured Representations for Video Understanding
• 2016 - ACM Multimedia web chair

Recent talks

• 2024 - Lecturer at Oxford Machine Learning Summer School, UK
• 2024 - Keynote at 1st Workshop on Human-Centered Vision and Media Technologies, Japan.
• 2024 - Keynote at AIST computer vision challenges workshop, Japan.
• 2023 - Keynote at CVPR 2023 anti-UAV workshop, Canada
• 2023 - Talk at eScience Center, the Netherlands
• 2023 - Talk at Science Park Imaging Symposium, the Netherlands
• 2023 - Talk at FruitPunch AI, the Netherlands
• 2023 - Talk at Rising Star Symposium at KAUST, Saudi Arabia
• 2022 - Lecturer at School in AI summer school, Italy
• 2022 - Talk at University of Bristol, UK. Host: prof. Dima Damen
• 2022 - Talk at Columbia University, USA. Host: dr. Carl Vondrick
• 2022 - Talk at FAIR, USA. Host: dr. Karen Ullrich
• 2022 - Talk at Utrecht University, the Netherlands. Host: dr. Ronald Poppe
• 2022 - Talk at Delft University of Technology, the Netherlands. Host: dr. Jan van Gemert

Grants

• PI - ELLIS PhD award with prof. Rita Cucchiara
• PI - NWO ClickNL
• PI - Google Perception Academic Funding with dr. Thomas Mensink
• PI - Data Science Center grant with dr. Erwin Berkhout
• PI - Informatics Institute collaboration grant with prof. Paul Groth
• co-PI - HAVA-Lab with prof. Cees Snoek
• co-PI - NWO ClickNL with dr. Nanne van Noord
• co-PI - ACTA AiO competition with dr. Erwin Berkhout

Awards

• Doctoral Consortium Award - ACM Multimedia 2016
• Benchmark winner - TRECVID Multimedia Event Detection 2015
• Reviewer award - CVPR21, ICLR21, ECCV20, ICML20, NeurIPS19, ICMR18, ICCV17, ICMR15

Teaching

• Applied Machine Learning 2018-now
• Education Committee MSc Information Studies 2019-now
• Thesis AI coordinator 2019-2022
• Entry committee MSc Artificial Intelligence 2019-2021
• Guest lecturer in Computer Vision I and Computer Vision II 2018-2021

Hyperspherical Prototype Networks

Code and pre-computed prototypes are available [here].

Shuffled ImageNet-Banks for Video Event Detection and Search

Pre-trained models can be downloaded by following the instructions [here].

Spatial-Aware Object Embeddings for Zero-Shot Localization and Classification of Actions

Code and pre-computed prototypes are available [here].

Water Detection through Spatio-Temporal Invariant Descriptors

The complete Video Water Dataset is available [here] (warning 14GB).