
Postdoctoral researcher @Inria_Saclay (@InriaMind team) in machine learning.
France
Joined July 2016
- Tweets415
- Following418
- Followers223
- Likes4.7K
Antoine Collas retweeted
Have you ever typed import mne in #Python?
Did it help you publish a paper, save hours, or even land your dream job? 💡
Now’s your chance to give back — @mne_news can finally receive 💸 donations to keep the lights on (CIs, forum, etc.) 🙌 github.com/sponsors/mne-tool…
Antoine Collas retweeted
I am thrilled to announce that our work on the generalization of flow matching has been accepted to NeurIPS as an oral!!
See you in San Diego 😎
New paper on the generalization of Flow Matching arxiv.org/abs/2506.03719
🤯 Why does flow matching generalize? Did you know that the flow matching target you're trying to learn **can only generate training points**?
with @Qu3ntinB, Anne Gagneux & Rémi Emonet 👇👇👇
Antoine Collas retweeted
🥳Very happy to see SKADA-Bench accepted at TMLR 2025 @TmlrOrg! This was a huge team effort to design a reproducible open-source Domain Adaptation benchmark with realistic validation.
📜 openreview.net/pdf?id=k9F63D…
🖥️ github.com/scikit-adaptation…
🧵 bsky.app/profile/rflamary.bs…
1/2
MNE-Python 1.10 now includes our EEG montage interpolation method from our EUSIPCO 2024 paper 🎉
→ .interpolate_to() works on Epochs, Evoked, ...
→ One line to align signals across caps
📄 Paper: arxiv.org/abs/2403.15415
🧠 Demo: mne.tools/stable/auto_exampl…
#EEG #MNEPython #AI
Antoine Collas retweeted
Also, many of my students and colleagues working in the field of AI feel that there is an unfair competition between the academia and the industry. I do not think so. While industry focuses on the technology, academia should focus on the methodology. I believe that, methodology-wise, current practice of AI does not have any major secrets and, in fact, it is rather primitive and brute force. That is the reason why it is easily emulated given sufficient resource for reverse engineering. This is another reason why closed-door proprietary practice is doomed to be surpassed.
Antoine Collas retweeted
Merci @lemondefr pour un joli résumé de mes aventures scientifiques et logiciels 📈📠
lemonde.fr/sciences/article/…
Beaucoup de messages qui me tiennent à cœur : travail d'équipe, logiciel libre, rigueur scientifique
Merci aux collègues et amis qui ont témoigné, je suis ému de lire
Super proud of this work! Domain Adaptation is the way to go for many applications and Skada will democratize it!
This Post is from an account that no longer exists. Learn more
Next week, we’ll present our spotlight paper at #NeurIPS2024 on domain adaptation for EEG data. Join us in East Exhibit Hall A-C on Friday at 4:30 PM! Learn how we address shifts in both data and label distributions. Looking forward to your thoughts!
This Post is from an account that no longer exists. Learn more
Antoine Collas retweeted
Geomstats first provides common manifolds embedded in vector spaces.
🌐 Symmetric Positive Definite (SPD) matrices, and its many families of Riemannian metrics
🌐 Hyperbolic spaces
🌐 Hyperspheres
🌐 Stiefel manifolds
🌐 Grassmanians
You can also learn your manifold from data!
Antoine Collas retweeted
Geomstats 2.8.0 is out: Discover differential geometry for learning in Python!
🌐 30+ manifolds
🌐 Connections & Riemannian metrics: geodesics, exp & log maps, parallel transport, curvatures
🌐 Lie groups, fiber bundles, quotient spaces, stratified spaces
github.com/geomstats/geomsta…
Antoine Collas retweeted
I am excited! 🎸🚀 Checkout our latest work accepted @NeurIPSConf on Riemannian domain adaptation for #EEG —led by @apmellot & @AntoineCollas — they prepared a very nice #thread 🪡 👇👇🏼👇🏿
This Post is from an account that no longer exists. Learn more
🚀I’m beyond excited to share that our paper has been accepted as a Spotlight at #NeurIPS2024! Super proud of the team’s great work! Once again, we make learning across neuroscience datasets easier!
This Post is from an account that no longer exists. Learn more
Antoine Collas retweeted
99% ML engineer i talked to don't realize gradient is actually defined via inner product. If you have a different definition of inner product (metric), you have different definition of gradient. This will free up your mind.. now preconditioning and layerwise lr tuning makes lot more sense: its actually redefining your notion of distance.
Antoine Collas retweeted
🤗Officially started Ph.D. with @IevgenRedko, @rtavenar and Laetitia Chapel @Inria @irisa_lab on Transformers & Distribution Shifts
🥳🇨🇦 Also, 2 papers accepted at #NeurIPS2024
📈 *Spotlight* arxiv.org/pdf/2406.10327
✋🏾 MaNO arxiv.org/pdf/2405.18979
More details soon!
Antoine Collas retweeted
#NeurIPS2024 decisions are out
Pymanopt 2.2.1 is here!
Highlights include bug fixes, support for SciPy versions >=1.13, and updated documentation. 🎉 Special thanks to new contributors! See: github.com/pymanopt/pymanopt…
#Python #Optimization #Manifolds
Antoine Collas retweeted
Super happy to announce the release of TorchDR 0.1 : the Dimensionality Reduction (DR) library using @PyTorch. Joint work with an amazing team : @RFlamary , @mathusmassias, @nicolas_courty, @CedricCuaz, Titouan Vayer, Aurélien Garivier. Thread 🧵.
Repo : github.com/TorchDR/TorchDR
Antoine Collas retweeted
Geomstats is a collaborative effort led by Luis Pereira, Alice Le Brigant, Xavier Pennec, Nicolas Guigui, Jules Deschamps, Adele Myers, Antoine Collas, Anna Calissano, Pablo Suarez-Serrato, Francisco Acosta, Christian Shewmake, Sophia Sanborn, and many more!
@AntoineCollas @jujuzor @AdeleMyersPhD @hopfbifurcator @cashewmake2 @naturecomputes @geometric_intel






















