Vincent NGUYEN

Since 09/2021, I have been an associate professor at l’IUT / University of Orléans, France and a member of the research team CA (Constraints and Learning) of LIFO. From 2023 to 2025, I served as Director of Studies for second-year Computer Science students at IUT Orléans.

My research focuses on learning from limited supervision and adapting machine learning models to changing conditions. I am particularly interested in how learned representations evolve when models are updated, and how these changes affect their performance and reliability.

I study these questions in different settings, including continual learning and federated learning, with a focus on:

• Learning with limited labels: weakly supervised, semi-supervised, self-supervised, and positive-unlabeled learning.
• Representation learning and adaptation: continual and incremental learning, representation drift, and adaptation to new tasks and data distributions.
• Robustness and security: model reliability under distribution shifts, including backdoor attacks and defenses in federated learning.

Research Grants:

• 2026-2029 : ANR EnACA project on AI-based multimodal understanding of comics, manga, and related art forms. The research focuses on generalizable instance segmentation under visual style and domain shift (PI).
• 2023-2026: grant CIFRE with IMPACT. (Co-PI)
• 2023-2026: APR Optimedias (Artificial intelligence for the benefit of precision medicine in neuropsychiatry and infectiology) funded by the Region Centre-Val-de-Loire, France.
• 2022-2025 : ANR ASTRID ROV-Chasseur on weakly supervised learning and efficient deep models for robotic vision. (PI)
• 2022-2023 : ACHS: a scientific collaboration projet of the French Embassy in Vietnam on medical data analysis (PI)
• 2015-2016 : eDuc: a technology transfer project funded by the SATT Grand Center on lecture videos indexing (PI)

Postdoctoral Researchers

• Hoàng-Ân Lê, postdoctoral researcher, ANR EnACA project, 2026–2028.

PhD Students

• Alexandre Jaud, PhD student in the ANR EnACA project, co-supervised with Jean-Christophe Burie (L3i, La Rochelle Université).
• Jean-Daniel De-Ambrogi, “Efficient Deep Models for Indoor Reconstruction and Navigation on Mobile Devices”, co-supervisor with Aladine Chatouni
• Trung-Anh Dang, “Continual learning”, co-supervisor with Christel Vrain
• Yen Vu, “Semi- and self-supervised learning with clustering using knowledge. Application to mineral potential mapping”, co-supervisor with Christel Vrain and Bich Dao and Hugo Breillard - BRGM
• Huy Thong Bui, bourse France Excellence Eiffel, “Federated learning”, co-supervisor with Son VU

Former Postdoctoral Researchers

• Etienne LEHEMBRE, postdoc of the project JUNON, 2025-2026.
• Rim Rahali, postdoc of the project ROV-Chasseur, 2023-2025.

---

past announcements:

• 10/2026: Our paper on rehearsal-free continual learning with a single shared prompt is accepted at WACV 2027.
  

• 09/2026: The University of Orléans and the Posts and Telecommunications Institute of Technology (PTIT, Vietnam) signed an academic cooperation agreement on September 11, 2026. I helped initiate and coordinate this partnership to foster joint research, academic exchanges, doctoral co-supervision, and the development of a potential double-degree Master’s program.
  

• 09/2026: Our paper CGS-SLAM: Collaborative Gaussian Splatting based SLAM for Multi-Agent Reconstruction was presented at the 3rd Workshop on Neural SLAM (NeuSLAM), ECCV 2026.
  

• 05/2026: 1 article on PU learning is accepted at ECML PKDD 2026.
  

• 03/2026: 1 article on geometric and structure constraints for continual learning is accepted in Machine Learning journal.
  

• 01/2026: 1 article on knowledge distillation for lightweight LLMs in sentiment analysis is accepted in LREC 2026.
  

• 12/2025: 1 article on semi-supervised object detection accepted in the journal Multimedia Tools and Applications.
  

• 11/2025: Recipient of the RIPEC bonus for excellence in academic activities and professional commitment.
  

• 10/2025: Our research on water resources, particularly groundwater time-series prediction using physics-informed neural networks (PINNs), has appeared in 9e édition de la Réunion des Sciences de la Terre 2025 and in Congrès Eau et IA 2026
  

• 09/2025: Project ANR PRC EnACA: Looking for 2 interns in Computer Vision and Machine Learning, 1 experienced researcher on domain adaptation segmentation link.
  

• 08/2025: Our project ANR PRC EnACA is accepted.
  

• 02/2025: 1 article on efficient models accepted in the journal Pattern Analysis and Applications.
  

• 01/2025: 1 paper on semi-supervised learning accepted at VISAPP 2025.
  

• 11/2024: 1 paper on PU learning accepted at SAC 2025.
  

• 10/2024: 1 paper on Continual learning accepted at WACV 2024.