Scholar
Pascal Mettes
Google Scholar ID: sMQxA3AAAAAJ
Assistant Professor, University of Amsterdam
Citations & Impact
Citations
2,220
H-index
26
i10-index
43
Publications
20
Co-authors
21
list available
Publications
Resume (English only)
Academic Achievements
- - ICLR 2025: 'Compositional Entailment Learning for Hyperbolic Vision-Language Models' (oral)
- - CVPR 2025: 'Hyperbolic Safety-Aware Vision-Language Models' (spotlight)
- - TMLR 2024: 'Intriguing Properties of Hyperbolic Embeddings in Vision-Language Models'
- - ICML 2025: 'Low-distortion and GPU-compatible Tree Embeddings in Hyperbolic Space'
- - ESWC 2025: 'Designing Hierarchies for Optimal Hyperbolic Embedding' (best paper nomination)
- - IJCV 2025: 'SimZSL: Zero-Shot Learning Beyond a Pre-defined Semantic Embedding Space'
- - CVPR 2022: 'Hyperbolic Image Segmentation'
- - NeurIPS 2021: 'Hyperbolic busemann learning with ideal prototypes'
- - CVPR 2020: 'Searching for Actions on the Hyperbole'
- - IROS 2024: 'Hyp2Nav: Hyperbolic Planning and Curiosity for Crowd Navigation' (oral)
Research Experience
- Research areas include:
- - Vision-language models should be hyperbolic
- - Hierarchical deep learning with hyperbolic embeddings
- - Robust deep learning in hyperbolic space
Background
- Pascal Mettes is a tenured Assistant Professor at the University of Amsterdam. The mission of him and his team is to advance the field of hyperbolic deep learning. Currently, deep learning is centred around Euclidean geometry, which has critical blindspots, especially in handling hierarchies. Their research focuses on developing theories and algorithms to perform deep learning in hyperbolic geometry, the natural geometry of hierarchies.
