Introducing "Automatic Integration (AutoInt)": a new method using neural networks to learn closed-form solutions to integrals. computationalimaging.org/publications/a… We demonstrate AutoInt for neural volume rendering, achieving >10x speedup compared to NeRF! (1/n) Julien N.P. Martel Gordon Wetzstein

With a contribution on perception and representation learning by Michel Devy & Frédéric Lerasle LAAS-CNRS UT3 Paul Sabatier and @cfevotte & myself IRIToulouse Toulouse INP ANITI Toulouse.

Look what arrived.. physical copies of our article in PNAS on the frontier of simulation based inference Gilles Louppe and Johann Brehmer pnas.org/content/117/48…

In 2016, AlphaGo was introduced. Two years later, its successor - AlphaZero - showed significant progress in Go, chess and shogi. Today in nature, our team describes MuZero, a significant step forward in the pursuit of general-purpose algorithms: go.nature.com/38qb2I8

Fruitful (yet remote) meeting of the ANITI Toulouse #AI Research Chair on fusion-based inference from heterogeneous data with colleagues and Ph.D. students from IRIToulouse ISAE-SUPAERO Centre d'Etudes Spatiales de la Biosphère CNES NXP & #CS_Group. 🔗dobigeon.perso.enseeiht.fr/aniti/

23 juin, 17h30 | Afterwork ANITI Toulouse avec Mathieu Fauvel, Coeurdoux Florentin, Y. Zerah et D. Salle NXP : "Apprendre avec peu de données ou à partir de données complexes". On parlera #IA, #DeepLearning et modèles génératifs. aniti.univ-toulouse.fr/2022/06/14/aft…

📄New accepted paper to be presented at ECML PKDD next september in Grenoble, France! Joint work with Coeurdoux Florentin from IRIToulouse/Toulouse INP & ANITI Toulouse and P. Chainais from CRIStAL/@Centralelille. 🔗arxiv.org/abs/2207.01246

📄New accepted paper to be presented at ESANN next october in Bruges, Belgium! Joint work with Coeurdoux Florentin from IRIToulouse/Toulouse INP & ANITI Toulouse and P. Chainais from CRIStAL/@Centralelille. 🔗arxiv.org/abs/2207.05468

David Picard Big thanks top Elsa Cazelle, Arthur Leclair, Rémi Cornillet, Cedric Vincent-Cuaz , Simon Mignon and Coeurdoux Florentin for their very nice presentations in the session.

📄 Un papier sur le transport optimal a été sélectionné à la conférence sur le Machine Learning ECML PKDD. C'est l’occasion de faire un focus sur le transport optimal en #informatique avec Nicolas Dobigeon et Coeurdoux Florentin de l'équipe SC - Dept. SI ↪️ irit.fr/un-papier-sur-…

Maximise expected info gain in model params (aka BALD) → effective active learning? Sadly not! EPIG, our alternative, shares BALD’s 1950s foundations but targets predictions. arxiv.org/abs/2304.08151 #AISTATS2023 w/ Andreas Kirsch 🇺🇦 Sebastian Farquhar Yarin Adam Foster Tom Rainforth 1/5

New #arXiv preprint introducing a stochastic counterpart of PnP-ADMM, embedding a #deep generative model (here a DDPM) as a implicit prior for Bayesian inference. Joint work w/ Coeurdoux Florentin (IRIToulouse, ANITI Toulouse) & P. Chainais (#CRIStAL). 🔗arxiv.org/abs/2304.11134

New #arXiv preprint introducing NF-SAILS, a sampling method for pre-trained normalizing flows. It exploits a Langevin dynamics formulated in the latent space. Joint work w/ Coeurdoux Florentin (IRIToulouse, ANITI Toulouse) & P. Chainais (#CRIStAL). 🔗arxiv.org/abs/2305.12149

Corrigendum: the co-authors are E. C. Faye and M. D. Fall, both from Institut Denis Poisson (IDP, Orléans). Sorry @FallMDiarra for the typo on your name! 😳

📄 Recently accepted for publication in IEEE Transactions on Image Processing. Again, congrats to my former Ph.D. student Coeurdoux Florentin for this work! 🎉 🔗doi.org/10.1109/TIP.20…

📄 Recently accepted for publication in Machine Learning. (last) Congrats to my former Ph.D. student Coeurdoux Florentin for this (last) work! 🎉 🔗link.springer.com/article/10.100… [OA]

Another open access #MLJ online-first paper today (or yesterday, actually): "Normalizing flow sampling with Langevin dynamics in the latent space" by Florentin Coeurdoux, Nicolas Dobigeon & Pierre Chainais (link.springer.com/article/10.100…) #OA