Maxime Haddouche

PhD Student at #InriaLondon. Some of my scientific interests: PAC-Bayes, online learning, optimisation.

[Maxime Haddouche @maxhaddouche]

@Yadkori It's really beautiful, congrats for this work !! Looks great to exploit some properties of the data distribution to avoid an explicit impact of the dimension

[Ambroise Odonnat @AmbroiseOdonnat]

🚀 We are thrilled to say that NeurIPS@Paris is back for a 5th edition to discuss recent advances in AI in the center of Paris!

🤗Registration: https://t.co/tI2K8J3m40
📅 25th & 26th Nov. 2025
🌐 https://t.co/tgZJ1h2Pyd

[Ambroise Odonnat @AmbroiseOdonnat]

🚀 NeurIPS@Paris is back for a 5th edition at the SCAI. Meet us in central Paris to discuss recent advances in AI!
📆 25th & 26th Nov. 2025
🌐https://t.co/tgZJ1h2hIF
🎓 Committee: Chloé-Agathe Azencott, @BachFrancis, Claire Boyer, @gerardbiau , @VianneyPerchet , @jeanphi_vert

[Maxime Haddouche @maxhaddouche]

@AntoninSchrab @ai_ucl @GatsbyUCL @bguedj @ArthurGretton Well done Antonin!!!

[Pierre Marion @PierreMari0n]

NeurIPS in Paris announcement: due to a large number of registrations that strongly exceeds the event’s capacity, we will close registration early on next Friday. Please register by then here: https://t.co/OpCAj1HJvP

[Maxime Haddouche @maxhaddouche]

@ArthurGretton @AntoninSchrab Congrats @AntoninSchrab !!!

[Maxime Haddouche @maxhaddouche]

A really nice event if you're in Europe in December and want to learn about recent advances in AI !

[Pierre Marion @PierreMari0n]

We are thrilled to say that NeurIPS@Paris is back for a 4th edition on the 4th and 5th of December 2024 at @Sorbonne_Univ_. A great occasion to meet + discuss recent advances in ML in central Paris!

More info: https://t.co/OpCAj1HJvP

Registration: https://t.co/PVQBMQfrbn

[Maxime Haddouche @maxhaddouche]

@mathusmassias @agramfort @lrntzrsc @RemiGribonval @RFlamary @olivier_cappe That's a really nice post, thanks a lot!!

[Maxime Haddouche @maxhaddouche]

@bremen79 Awesome, please don't hesitate if anyone in your team has any questions! :)

[Maxime Haddouche @maxhaddouche]

@bremen79 The question to find the best divergence (and lower bounds), could then be extended to find the best pair (IPM,divergence)!
Thanks again for this work, I love the idea of involving optimisation techniques for generalisation!

[Maxime Haddouche @maxhaddouche]

@bremen79 But those bounds suffers from weaknesses (no explicit convergence rates) then we recently focused (with co-authors) on mixing those two approaches to have the best-of-both worlds

https://t.co/N5yJdhHj6C

[Maxime Haddouche @maxhaddouche]

@HaghifamMahdi Hi Mahdi, thanks for your response.
That is intriguing indeed, will have a deeper look on this!

[Mathurin Massias @mathusmassias]

With @tvayer we're organizing a one day workshop on dimensionality reduction, November 10th at @ENSdeLyon : https://t.co/IBM0JP14wJ

We encourage short talks submissions from PhD students, and we can eventually fund their trip. RT appreciated!!!

[Maxime Haddouche @maxhaddouche]

Glad to see our joint work with @paulviallard @umutsimsekli and @bguedj accepted to #NeurIPS2023 ! Mixing Wasserstein distances with PAC-Bayes yield to tractable, generalisation-driven algorithms, not only for stochastic, but also deterministic predictors! https://t.co/DgcAWfzJf1

[Maxime Haddouche @maxhaddouche]

@BachFrancis Hope you'll still be recruiting in a few years !

[Maxime Haddouche @maxhaddouche]

@AntoninSchrab Congrats!!! Super happy for you :)

[Antonin Schrab @AntoninSchrab]

"The last MMD paper you'll ever need" is
now published at JMLR!

MMD Aggregated Two-Sample Test
https://t.co/6A5w2FlCoR

MMDAgg:
-Aggregation: Multiple Kernels
-Power: Minimax Sobolev Optimal
-Level: Non-Asymptotic
-Permutations & Wild Bootstrap
-JAX code:
https://t.co/o4tmHQaoE4