I am a Lecturer at Sorbonne University, affiliated with the Machine Learning – Deep Learning and Information Access (MLIA) team of the ISIR lab. My research focuses on machine learning, particularly in the areas of image understanding and multi-modal data integration, including text-image and multi-sensor inputs. I am particularly interested in developing hybrid models that combine both physical and statistical approaches. Additionally, I work on enhancing the robustness of machine learning models to ensure their reliability across diverse applications. Prior to joining Sorbonne University, I was an Assistant Professor at the CEDRIC laboratory of the Conservatoire National des Arts et Métiers (CNAM), where I was part of the Vertigo Team.
Sorbonne University – Campus Pierre et Marie Curie
Pyramid level 0 between towers 55 and 56
4, place Jussieu, 75005 Paris, France
----------------------------------
Office H13
clement.rambour@isir.upmc.fr
12/09/2024
Our paper on few-shot learning for Vision-Language Models (VLM) has been accepted at ECCV24, demonstrating the best accuracy-robustness trade-off.
10/07/2024
Thrilled to announce that my young researcher project "MERCATO: Multimodal Uncertainty Quantification for Earth Observation" has been funded by the French National Research Agency! 🙌
20/03/2024
PhD defense of E. Ramzi: Robust image retrieval with deep learning! 🙌 🥳
21/12/2023
PhD defense of L. Themyr: 3D medical image segmentation! 🙌 🥳
25/04/2023
Accepted paper at ICML23 on hybrid EBM models for OOD detection
20/11/2022
Accepted internal laboratory project ORION on the design of robust models
18/09/2022
Accepted paper at WACV23 on transformers for high resolution semantic segmentation
20/09/2022
Accepted paper at MLMI22 MICCAI workshop
04/07/2022
2 accepted papers at ECCV22 !
15/12/2021
We are hiring a postdoc to work on IA and inverse Problems. The job is expected to start in spring 2022. See the job description here FILLED
05/12/2021
We are lookink for an internship to work on RL and physical models. See the job description here (in french) FILLED
28/09/2021
Paper on robust and decomposable average precision accepted at NeurIPS'21
20/09/2021
Paper U-Net Transformer accpeted at MLMI
26/07/2021
ANR Astral on machine learning for multidimensional SAR data processing accepted !
30/09/2020
Presentation of our work on SAR tomography for urban areas at the ALYS workshop
01/10/2020
I will present our work on multi-modal (SAR and optical Sentinel images) series analysis at the ESA Phi-week virtual event