Thibaud Rahier

Math & mountain lover. PhD in Machine Learning, Researcher @CriteoAILab

[Criteo AI Lab @CriteoAILab]

We are thrilled to announce this deserved award at #ALT25.

Congrats to Marc Abeille, David Janz, @CiaraPikeBurke 👏

Paper: When and why randomised exploration works (in linear bandits)
https://t.co/Ovfgs37t7F

[Thibaud Rahier @RahierThibaud]

Very proud to be your advisor Julien, this is very solid work!

[Julien Zhou @jlnzhou]

Glad to announce that my first paper got accepted to #NeurIPS24 ! Yay!
"Towards Efficient and Optimal Covariance-Adaptive Algorithms for Combinatorial Semi-Bandits" (https://t.co/N8dPKm2DTY) 1/4

[Julien Zhou @jlnzhou]

Glad to announce that my first paper got accepted to #NeurIPS24 ! Yay!
"Towards Efficient and Optimal Covariance-Adaptive Algorithms for Combinatorial Semi-Bandits" (https://t.co/N8dPKm2DTY) 1/4

[Julyan Arbel @JulyanArbel]

🚀 Covariance-adaptive algorithms for combinatorial semi-bandit feedback with awesome bounds!

Congrats @jlnzhou on your first accepted paper! #NeurIPS2024 🥳🎉

[Alexandre Ramé @ramealexandre]

Introducing Weight Averaged Rewarded Policies (WARP), Google DeepMind's latest RLHF alignment method using the magic of model merging. By scaling alignment like pre-training was scaled, WARP learns sota Gemma LLM surpassing previous releases. A 🧵below. https://t.co/Ck2VWNQKBA

[Criteo AI Lab @CriteoAILab]

How #ChatGPT and #LLM revolutionized language processing? Discover the groundbreaking research led by @LivaRalaivola and Criteo's AIR team as we unravel the mysteries of these technologies and their implications for the future of AI-driven innovation.🔮
https://t.co/nA5vAaCVIK

[Masha Vladimirova @iamvladimirova]

Since I joined @CriteoAILab, I started digging into applied causality topics. Our recent work “Maximizing the Success Probability of Policy Allocations in Online Systems” got accepted to AAAI-24 🥳!
https://t.co/XFjziS9bEI
Ping @artembetley @RahierThibaud
🧵🧵🧵

[Alexandre Ramé @ramealexandre]

Exciting news 🎓! I'm defending my PhD on "Diverse & Efficient Ensembling of Deep Networks" tomorrow at 13h30 CEST. If you're in Paris and can join, DM me. Or catch the live stream on YouTube: https://t.co/WIzDwbnl3J. Wish me luck!

[Houssam Zenati @HoussamZenati]

Extremely happy that my latest PhD work on "Sequential Counterfactual Risk Minimization" has been accepted at #ICML2023 🎉🥳🎊

https://t.co/YfluXRt0HQ

Work at @CriteoAILab with Eustache Diemert, @mmatthieumartin, and @inria_grenoble with @julienmairal, Pierre Gaillard

[Heura Foods 🇫🇷 @HeurafoodsFr]

Voici un chiffre stupéfiant : 70 % de la population française croit que manger local est meilleur pour le climat que devenir végétarien.

C'est totalement faux mais savez-vous pourquoi?

[Alexandre Ramé @ramealexandre]

But we go a step further by averaging weights fine-tuned from different initializations. This requires new relaxed averageability conditions, which we analyze in our paper. Rough insight: if two initializations can be averaged, then so can their fine-tuned weights.

[Alexandre Ramé @ramealexandre]

Ready to give your deep models a second life? Introducing model ♻️ recycling (https://t.co/ReB2ruXxsY), improving generalization by reusing weights fine-tuned on various vision tasks. Just like you recycle your bottles and cardboards, it's time to start recycling your models too!

[Alexandre Ramé @ramealexandre]

Weight averaging strategies are super useful in deep learning, and succeed despite the non-linearities in networks' architectures. Our 2 works presented at #NeurIPS2022 analyze how they can help for out-of-distribution classification in computer vision! (1/4)

[MLIA @mlia_isir]

End of two fantastic research days at the Neurips@Paris workshop organized by #SCAI and ⁦@CriteoAILab⁩

Thanks for the organization!

[ECML PKDD @ECMLPKDD]

There is a winner at the discovery challenge Criteo AI organised! Congratulations and thank you Criteo for the successful organization! More information at Program/Discovery Challenges

[Liva Ralaivola @LivaRalaivola]

Latest publications from @CriteoAILab at @NeurIPSConf (x 7) and @TmlrOrg: https://t.co/rCeBN0ihWp

Congrats to all Criteos and co-authors for pushing ML. @m_kirchmeyer @RahierThibaud @rakotal1 @VianneyPerchet @MaximeVono @SergeyI49013776
@RLerallut

[Thibaud Rahier @RahierThibaud]

Feel free to join us at our @CriteoAILab workshop at @ECMLPKDD on out-of-distribution generalization at 2:30PM today, with an invited talk by @ramealexandre and a keynote by David Lopez-Paz from @MetaAI !

[Criteo AI Lab @CriteoAILab]

💻Join us for the Uplift Modeling Tutorial & Workshop @ECMLPKDD (ITE, CATE, HTE for business decision support)
📢Keynote by Eustache Diemert
🕹Discovery Challenge by Eustache, Karim Kassab and Thibaud Rahier
More details 👉 https://t.co/s1P4ZYaL23 #ecmlpkdd2022

[Criteo AI Lab @CriteoAILab]

💻Join us for the Uplift Modeling Tutorial & Workshop @ECMLPKDD (ITE, CATE, HTE for business decision support)
📢Keynote by Eustache Diemert
🕹Discovery Challenge by Eustache, Karim Kassab and Thibaud Rahier
More details 👉 https://t.co/s1P4ZYaL23 #ecmlpkdd2022

[Alexandre Ramé @ramealexandre]

Interested in linear mode connectivity, weight averaging and/or OOD generalization? Check out our latest DiWA paper, now accepted at #Neurips2022

[Houssam Zenati @HoussamZenati]

🎉Excited 🎉 to present our paper on nested bandits at #ICML2022 Join us at the📍Theory session room 307 at 🗓️11:35 EDT to attend our spotlight📣! We propose a nested bandit algorithm that performs a layered exploration when the set of alternatives enjoys a similarity structure.