RT @XReyRobert: Pour les inscriptions: . Pour l'abstract du papier de recherche de @EloiZablocki: .

Excited to share the first preprint from the @IBMResearch Paris team. We present a differentiable rule induction algorithm, integrating the interpretability of rule learning with a neural network based prediction model. @IBM_France

Last Friday I had the honour of seeing @GJ_Both succesfully defend his PhD. I am very grateful for all the wonderful work he did during the last three years. It was a pleasure working together.

RT @XReyRobert: 2ème session du Café -Séminaire IBM Research mensuel sur l'#IA interprétable ce Jeudi 2 à 14h30 !. Nous accueillons Le Pr.….

RT @RobertLustigMD: "We demonize certain drugs, and celebrate others," says Michael Pollan (@michaelpollan). "Sugar is a drug, and celebrat….

RT @stormkees: This is pure gold. Or germanium, who can tell. H/T .@jim_deane

RT @arXiv_Daily: Sparsistent Model Discovery.by Georges Tod et al. including @RemyKusters.#DeepLearning #Hyperparam….

RT @arXiv_Daily: Fully differentiable model discovery.by Gert-Jan Both and @RemyKusters.#NeuralNetwork #ComputerSci….

Our team at @IBMResearch is hiring a senior research scientist (Paris-Saclay) to work on safe, explainable and responsible AI systems. If you have 1+ years of research experience after PhD and are excited to join the team, get in touch!

After three fantastic years at @CriResearch I am happy to start a new adventure at @IBMResearch where I will continue to work on interpretable and explainable machine learning

RT @luyibov: W/@RemyKusters and @GJ_Both we worked on classification of trajectories based on a set of features of them inspired by the cha….

New paper is out on performing automated model discovery using DeepMoD when very few sensors are available to collect the data. Paper will be presented at the ICLR workshop on AI for Modeling Oceans and Climate Change, May 7th!.

The next round for research fellows @CriResearch is open again. Join this fantastic place to do research in the center of Paris!.

Interesting read on how AI research shifted from theoretical science to an experimental science but lacking a proper methodology. In my own experience, the majority of papers still don’t share their code/data resulting in an avoidable replication crisis:

RT @danielequercia: Trust - Our World in Data

Trying to discover interpretable models from data: Our team just released a modular package for deep learning-driven model discovery of PDEs and ODEs: DeePyMoD. Paper, code, and documentation is available on:.@GJ_Both @criparis.

@CULLYAntoine @matkomorowski @DuleFR @_Yann_ @NataliaDiazRodr @luyibov @danielequercia @fatimaWehbi @HubFranceIA @jgrizou @FrontiersIn @CriResearch .5/5.

3) AI education should receive more attention, efforts, and innovation from the educational and scientific communities. 4/5.

2) AI research must be accompanied by decision explainability, dataset bias transparency as well as the development of evaluation methodologies and the eation of regulatory agencies to ensure responsibility. 3/5.

We analyze three key challenges to interdisciplinary AI research and deliver three broad conclusions: .1) Future development of AI should not only impact other scientific domains but should also take inspiration and benefit from other fields of science. 2/5.