Hello, I'm Eva Feillet.

I am currently Assistant Professor (Maîtresse de Conférence en Informatique) at Paris-Saclay University. I am a member of the LISN laboratory (Laboratoire Interdisciplinaire des Sciences du Numérique > Sciences et Technologies des Langues > Sémantique et Extraction d'Information), and I teach at UFR Sciences d'Orsay.

My current research interests include :
- Resource-efficient training of deep neural networks, including continual learning
- Visual representation learning and semantic aspects in computer vision
- Multimodal learning, with a focus on vision and language


Previously, I was a postdoctoral researcher in frugal deep learning at Université Paris-Dauphine, where I worked with Pr. Alexandre Allauzen in the MILES team of LAMSADE laboratory.
During this postdoc, I was part of the SHARP research project, funded by the France 2030 program and managed by the ANR in the context of PEPR IA.

I did my PhD in continual learning at CEA-list / CentraleSupélec MICS in Paris area, France. During my PhD, I worked under the direction of Pr. Céline Hudelot (CentraleSupélec MICS). I was also supervised by Dr. Adrian Popescu (CEA-list) and Dr. Marina Reyboz (CEA-list).
I hold an engineering degree from CentraleSupélec Engineering school (equivalent to M.Sc. in AI).

Current affiliation : LISN - Laboratoire Interdisciplinaire des Sciences du Numérique, Université Paris-Saclay, France

Publications

ICLRw 2026

Colagrande, A., Caillon, P., Feillet, E., Allauzen, A.

In: AI & PDE Workshop @ ICLR 2026

[paper]

ICASSP 2026

Feillet, E., Whetten, R., Picard, D., Allauzen, A.
Polynomial mixing for efficient self-supervised speech recognition

In: Proceedings of the ICASSP 2026

[paper] [code]

ICCVw 2025

Colagrande, A., Caillon, P., Feillet, E., Allauzen, A.
Linear Attention with Global Context: A Multipole Attention Mechanism for Vision and Physics

In: ECLR Workshop @ ICCV 2025 (2nd Workshop on Efficient Computing under Limited Resources).

[paper] [code]

Thesis manuscript

Feillet, E.
Analysis and Recommendation Methods for Class-Incremental Learning

Université Paris-Saclay, 2024

[access on HAL]

WACV 2025

Feillet, E., Popescu, A., & Hudelot, C.
A Reality Check on Pre-training for Exemplar-free Class-Incremental Learning

In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision. 2025. p 7614-7625.

[paper] [supp]

WACV 2025

Pegeot, T., Feillet, E., Popescu A., Kucher I., & Delezoïde,s B.
Temporal Dynamics in Visual Data : Analyzing the Impact of Time on Classification Accuracy

In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision. 2025. p 6932-6943.

[paper] [supp] [data]

ICPR 2024

Feillet, E., Popescu, A., & Hudelot, C.
Recommendation of data-free class-incremental learning algorithms by simulating future data

[paper][supp]

In: Proceedings of the International Conference on Pattern Recognition. 2024.

WACV 2024

Feillet, E., Petit, G., Soumm M., Popescu, Delezoïde, B., Picard, D., & Hudelot, C.
An Analysis of Initial Traning Strategies for Exemplar-free Class-Incremental Learning

In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision. 2024. p. 1837-1847.

[paper] [supp]

WACV 2023

Feillet, E., Petit, G., Popescu, A., Reyboz, M., & Hudelot, C.
AdvisIL - A Class-Incremental Learning Advisor

In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision. 2023. p. 2400-2409.

[paper] [supp] [code]

RFIAP 2022

Feillet, E., Petit, G., Popescu, A., Reyboz, M., & Hudelot, C.
Incremental Learning under Memory Constraints: a scaling heuristic for deep convolutional models

In : Reconnaissance des Formes, Image, Apprentissage et Perception. 2022.

[paper] [code]

Teaching at Université Paris-Saclay

Currently, I teach the following courses at Université Paris-Saclay :
- Math for Data Science (M1, lectures and tutorials)
- Computer vision (M2, lectures, tutorials and projects), based on the course by Céline Hudelot and Maria Vakalopoulou (MICS)
- Frugal learning (M2, projects), with Adrian Popescu (CEA list)
- Introduction to machine learning (L3, tutorials), taught by François Landes (LISN)
- Introduction to data science (L1, tutorials), taught by Fanny Pouyet (LISN)
- AI research project supervision (TER, M1)

Previsously, I have been a teaching assistant at CentraleSupelec for the following courses:
- Deep learning course (Master 2), taught by Hervé Le Borgne (CEA list)
- Information systems and Programming (L3/Bachelor), taught by Gianluca Quercini (LISN)
- Coding weeks challenge (L1/Bachelor), taught by Paul Tourniaire (joint Bachelor CentraleSupélec + McGill University)

Participation to events

Journée ''Jeunes Statisticiens et Probabilistes'' (SFdS)

Introduction à l'apprentissage multimodal
Oratrice invitée par le groupe Jeunes de la Société Française de Statistiques
Institut Henri Poincaré, Paris, January 2026

Thesis committee

Member of Antoine Montmaur's thesis jury,
PhD directed by Pr. Ngoc Son Vu at ETIS Lab
Cergy (France), December 2025

JDSE 2021

Junior Conference on Data Science and Engineering.
Gif-sur-Yvette (France), September 2021.
Best student paper.

Job offers


[PhD] Contact me if you are interested in pursuing a PhD on
Small language models: efficient adaptation & continual evaluation;
Continual learning in the era of multimodal foundation models


[M2] Research internships 2026 at LISN lab / Positions filled !
Topic 1: Multimodal continual learning.
Topic 2: Benchmarking Parameter-Efficient Fine-tuning methods for Spoken Language Understanding.
Duration: 5 to 6 months (starting March 2026).
Location: LISN lab, Université Paris-Saclay, Orsay, France.
Supervisors: Eva Feillet and Sahar Ghannay.

Get in touch

Email: name [dot] surname [at] universite-p#ris-s#clay.fr