About me

My name is Laetitia Chapel and I am currently a full professor in computer science at Institut Agro Rennes-Angers. I lead the TAGADA research team of IRISA, whose aim is to develop a mathematical and algorithmic theory of geometric alignment in artificial intelligence. My main research topic is machine learning through alignment with a particular focus on structured data such as time series, graphs, hierarchical representations. I defended my HDR ("habilitation à diriger des recherches") entitled "machine learning for structured data" in 2022.

News

July 2026. Paper Sliced Transport Plans, with Eloi Tanguy and Julie Delon, has been accepted for publication in Mathematics of Computation.
May 2026. New preprint out Vision Transformer Finetuning Benefits from Non-Smooth Components, poster at ICML 2026.
March 2026. NeurIPS is going to take place in 3 locations this year: Sydney (main event), Atlanta and Paris. Stay tuned on https://neurips.cc/ for last updates. Interested in sponsoring the event? All the details here: here or contact sponsoring@neurips-europe.cc
Dec. 2025. Project MacLEOD (PEPR IA) has been accepted. Looking forward to continue working on Machine Learning on geOmetries and Distributions!
Oct./Nov 2025. I gave some talks about sliced Wasserstein Plans with application to CFM at AgroParisTech, Imaging in Paris seminar, ML at Aussois autumn school. You can download the slides here, here and here.
Aug. 2025. New preprint out Sliced Optimal Transport Plans.
May. 2025. New preprint out Bridging Arbitrary and Tree Metrics via Differentiable Gromov Hyperbolicity. Poster at NeurIPS.
May. 2025. New preprint out Differentiable Generalized Sliced Wasserstein Plans. Poster at NeurIPS.
Feb. 2025. Giving a keynote at Journées Transport optimal et ses applications en machine learning et analyse de données du GDR IASIS. The slides are on the Journées' website.
Feb. 2025. Paper 'Leveraging covariance representations with Riemannian ResNet for Satellite Image Time Series classification' has been accepted at IGARSS. This is a joint work with Khizer Zakir, Charlotte Pelletier and Nicolas Courty.
Jan. 2025. Paper One for all and all for one: Efficient computation of partial Wasserstein distances on the line has been accepted to ICLR

Experience

Current position

Teaching within Institut Agro Rennes-Angers.
Research within TAGADA research team from IRISA.

Short Vitae

Since 2023. Full professor at Institut Agro Rennes-Angers.
2010-2023. Associate professor (Maître de Conférences) at Université de Bretagne Sud.
2008-2010. Research fellow at the Hamilton institute of the National University of Ireland Maynooth.
2007. Ph.D. degree in computer science from Université Blaise Pascal, Clermont-Ferrand, France.

Students

I am (or was) implied in the advising process of several master students and of the following PhD thesis and post-docs:

Ambroise Odonnat (PhD started in 2024, with Huawei), Léo Buecher (PhD started in 2024), supervised with Nicolas Courty and Lucas Drumetz, Sergio Osson (PhD started in 2025), with Charlotte Pelletier, Guillaume Mahey, (phD 2024, supervised with Gilles Gasso), Manal Hamzaoui, (PhD 2023, supervised with Sébastien Lefèvre and Minh-Tan Pham), François Painblanc (supervised with Romain Tavenard and Chloé Friguet), Titouan Vayer (PhD 2020, supervised with Nicolas Courty and Romain Tavenard, now researcher at INRIA), Pierre Gloaguen, (post doc 2017, now assistant prof. at UBS), Adeline Bailly (PhD 2018, supervised with Romain Tavenard, now researcher at DGA), Yanwei Cui, (PhD 2017, supervised with Sébastien Lefèvre, now data scientist at covea).

Recent publications

(a more up-to-date list is available on [scholar] )

In chronological order.
[PROC]	Multi-Prototype Hyperbolic Learning Guided by Class Hierarchy (Paul Berg, Léo Buecher, Bjorn Michele, Minh Tan Pham, Laetitia Chapel, Nicolas Courty), International Journal of Computer Vision, 2025, [Springer].
[PROC]	One for all and all for one: Efficient computation of partial Wasserstein distances on the line (Laetitia Chapel, Romain Tavenard), ICLR, 2025, [OpenReview].
[JOURNAL]	Hyperbolic Prototypical Network for Few Shot Remote Sensing Scene Classification (Manal Hamzaoui, Laetitia Chapel, Minh-Tan Pham, Sébastien Lefèvre), Pattern Recognition Letters, 2024, [elsevier].
[PROC]	Horospherical Learning with Smart Prototypes (Paul Berg, Bjorn Michele, Minh-Tan Pham, Laetitia Chapel, Nicolas Courty), British Machine Vision Conference (BMVC), 2024, [Hal].
[PROC]	Fast Optimal Transport through Sliced Generalized Wasserstein Geodesics (Guillaume Mahey, Laetitia Chapel, Gilles Gasso, Clément Bonet, Nicolas Courty), NeurIPS, 2023, [Arxiv], Spotlight (approx. top 3% of the papers).
[PROC]	Match-And-Deform: Time Series Domain Adaptation through Optimal Transport and Temporal Alignmentn (François Painblanc, Laetitia Chapel, Nicolas Courty, Chloé Friguet, Charlotte Pelletier, Romain Tavenard), ECML PKDD, 2023, [Arxiv].
[PROC]	Hyperbolic variational auto-encoder for remote sensing scene embeddings (Manal Hamzaoui, Laetitia Chapel, Minh-Tan Pham, Sébastien Lefèvre), IGARSS, 2023, [HAL].
[JOURNAL]	Time series alignment with global invariances (Titouan Vayer, Laetitia Chapel, Nicolas Courty, Rémi Flamary, Romain Tavenard, Yann Soulard), TMLR, 2022 [pdf]
[REPORT]	Hyperbolic Sliced-Wasserstein via Geodesic and Horospherical Projections (Clément Bonnet, Laetitia Chapel, Lucas Drumetz, Nicolas Courty), 2022 [Arxiv]
[PROC]	A hierarchical prototypical network for few-shot remote sensing scene classification (Manal Hamzaoui, Laetitia Chapel, Minh-Tan Pham, Sébastien Lefèvre), ICPRAI: Pattern Recognition and Artificial Intelligence, Lecture Notes in Computer Science, vol 13364., 2022. [pdf] Best paper award.
[PROC]	Unbalanced Optimal Transport through Non-negative Penalized Linear Regression (Laetitia Chapel, Rémi Flamary, Haoran Wu, Cédric Févotte, Gilles Gasso), NeurIPS, 2021 [Arxiv](* indicates equal contribution)
[JOURNAL]	Scalable clustering of segmented trajectories within a continuous time framework: application to maritime traffic data (Pierre Gloaguen, Laetitia Chapel, Chloé Friguet, Romain Tavenard), Machine learning, 2021. [HAL]
[JOURNAL]	POT: Python Optimal Transport (Rémi Flamary and Nicolas Courty and Alexandre Gramfort and Mokhtar Z. Alaya and Aurélie Boisbunon and Stanislas Chambon and Laetitia Chapel et al.), Journal of Machine Learning Research, 22(78), pages 1-8, 2021. [pdf]
[PROC]	Partial Optimal Transport with Application on Positive-Unlabeled Learning (Laetitia Chapel, Mokhtar Z. Alaya, Gilles Gasso), NeurIPS, 2020 [pdf]
[JOURNAL]	Fused Gromov-Wasserstein distance for structured objects (Titouan Vayer, Laetitia Chapel, Rémi Flamary, Romain Tavenard, Nicolas Courty), Algorithms, special issue Efficient Graph Algorithms in Machine Learning, 2020.[doi]
[PROC]	Optimal transport for structured data with applications on graphs (Titouan Vayer, Laetitia Chapel, Rémi Flamary, Romain Tavenard, Nicolas Courty), 36th International Conference on Machine Learning, PMLR 97:6275-6284, 2019.[pdf]
[PROC]	Sliced Gromov-Wasserstein (Titouan Vayer, Rémi Flamary, Romain Tavenard, Laetitia Chapel, Nicolas Courty), Advances in Neural Information Processing Systems 32 (NeurIPS), 2019.[pdf]

Contact

Agrandir le plan

Addresses

Laboratoire IRISA
Campus de Tohannic
56000 Vannes, France
Office C113
Phone : (+33)2 97 01 72 51
Fax : (+33)2 97 01 72 79

L'Institut Agro Rennes-Angers
65, rue de Saint-Brieuc
35000 Rennes, France.

In order to reach me

firstname dot lastname AT irisa.fr

_____________________________________________________
L'Institut Agro Rennes-Angers / IRISA / Obelix team - Last update: 05/2026