Remy Sun

Remy Sun

Research scientist

Université Côte d'Azur / Inria / I3S

About me

I am a research scientist (Inria Starting Faculty Position) in the MAASAI team at Inria d’Université Côte-d’Azur since October 2024. My current research interests focus mostly on knowledge injection in neural networks, multimedia interpretation, deep learning for physics and conditional/guided diffusion models. I was previously a postdoctoral researcher in this same MAASAI team at Inria Sophia Antipolis (advised by Diane Lingrand and Frederic Precioso) working on the MultiTrans project. I previously defended my PhD in late 2022 on content combination strategies for image classification at Sorbonne University’s Machine Learning and Information Access (MLIA) team and Thales Land and Air Systems under the supervision of Matthieu Cord, Nicolas Thome, Clément Masson and Gilles Hénaff.

Before that, I worked as a research intern at Conservatoire National des Arts et Métiers on deep learning for EEG classification, at the Empirical Inference department of Tuebingen’s MPI for Intelligent Systems on causal analysis of deep generative models, at IST Austria on detection of OOD network predictions and at IRISA on deep representations for protein sequences.

Recent & Upcoming News

  • May 18-22, 2026 We are holding a summer school at UniCa on Machine learning methods for physical simulations (data-driven, pinns, deep reinforcement learning, bayesion optimization)!
  • April 13, 2026 Lakshmi Manne (co-advised with Frédéric Precioso) started as an intern on knowledge injection for Visual-Language-Action models. Welcome Lakshmi!
  • April 13, 2026 Muhammad Arslan (co-advised with Diane Lingrand) started as an engineer in the team on few shot classification methods. Welcome Muhammad!
  • March 3, 2026 We have a new preprint on arxiv Beyond Mixtures and Products for Ensemble Aggregation: A Likelihood Perspective on Generalized Means
  • February 25, 2026 We have a new preprint on HAL Leveraging Computer Vision for Efficient and Scalable Biodiversity Monitoring in Marine Ecosystems: A Multi-Year Study on 3 Ecologically Important Fishes
  • February 12, 2026 Our paper Automated Counting of Fish in moving Diver Operated Videos (DOV) for Biodiversity Assessments was accepted at Methods in Ecology and evolution Congrats Kilian!
  • January 13, 2026 Our paper When Are Two Scores Better Than One? Investigating Ensembles of Diffusion Models was accepted at TMLR Congrats Raphael!
  • December 1, 2025 Juliette Girardin started her intenship on improving physics inspired neural networks with existing knowledge Welcome Juliette!

Contact information