"AI amplifies whatever is already there. Good discipline becomes great output. No discipline becomes technical debt at machine speed. Anthropic chose a direction. Go faster. Have Claude check Claude. And when it breaks, go faster still." substack.com/home/post/p-...
The team of JupyterCon 2023, PyData Paris 2024 & 2025 organizes a new conference named Compute! Paris 2026 on open source computation and data. The event will take place on November 25–26, 2026 at Sorbonne Université in Paris.
CfP deadline: May 24, 2026: compute.events/paris2026/cf...
Elle s'intéresse peut-être entre autre au langage utilisé
Et donc se reconverti au pipo
Oui, c'est une stratégie de management incarnée par la chaîne de commandement.
Les ZRRs le ministère de l'intérieur les veut. On peut lire ça comme une guerre entre l'intérieur et la la recherche.
Les HDRs, c'est beaucoup créé et subit par la recherche.
Ne pas s'imaginer qu'une administration, ou un état, c'est un ensemble cohérent, mais plutôt une juxtaposition d'intérêts.
Oh, et un côté influence, avec les entreprises meilleurs pour souffler les idées de comment il faut dépenser
Possible que ta remarque s'applique plus à cet environnement.
Les lourdeurs administratives sont des stratégies de pouvoir de la part de plusieurs niveaux hiérarchiques, y compris des échelons intermédiaires.
Ces tendances sont naturelles dans toute structure mais nous avons un manque de travail pour le combattre de la part d'au dessus.
Mon expérience à côtoyer le gouverneur me fait dire que ce n'est pas que la classe politique déteste les universitaires, mais elle fait systématiquement ses arbitrages en faveur de l'industrie.
Perles du Sénat. Le ministre de l'ESRE :
« je veux rappeler (...) qu'il n'y a là aucune discrimination politique [dans les refus d'embauche en ZRR]. La preuve en est que l'immense majorité des ZRR concerne les sciences dites dures, en particulier les technologies. »
www.senat.fr/questions/ba...
New skrub release ✨️
I'am really excited about the more general ApplyToCols.
I've found that it enables me to write very naturally complex data transformations on dataframes, as I combine it with skrub's selectors to choose which columns I apply transformations on.
skrub-data.org/stable/refer...
The minimum required version of polars has been increased from 0.20 to 1.5.
The TableReport custom filters have been improved and expanded: they can now take skrub selectors for filtering columns. The interface has also been simplified.
The has_nulls selector can now select columns based on a user-specified threshold of null values.
It is now possible to provide custom null values to the Cleaner, so that they are marked as nulls (for example, the string "unknown").
The performance of DataOps with many computational nodes has been improved. Additionally, DataOps CV splitters can now take kwargs. For example, this allows to specify groups when creating train/test splits.
The SingleColumnTransformer and RejectColumn classes allow the construction of custom-made transformers for specific use cases.
The ApplyToCols transformer is now a powerful alternative to the regular scikit-learn ColumnTransformer. It is now possible to apply any transformer to a subset of chosen columns using the skrub selectors.
✨ skrub version 0.8.0 has been released ✨
This version includes several new features, including multiple improvements to the functionality and performance of the Data Ops, along with a few bug fixes and improvements to the docs.
Changelog:
skrub-data.org/stable/CHANG...
Highlights below ⤵️
Try skrub skrub-data.org for machine learning with dataframes, and skore, still young but aiming to help evaluation and tracking of data-science docs.skore.probabl.ai
Both by creators of scikit-learn
Claude should be using them, but doesn't.
Today NeurIPS is announcing our official satellite event in Paris.
After responding to the call from Ellis following the success of EurIPS in December, we are pleased to reach a new milestone by joining forces with the NeurIPS organizing committee for the 2026 edition.
The state of the art for learning or tabular data is:
arxiv.org/abs/2602.11139
It comes with a high-quality software implementation:
tabicl.readthedocs.io/en/latest/
💰 €1M award pool for Medical Imaging AGI’s Last Exam (MEDAL)
Medical AI papers are booming - but are we solving the right problems?
Too often, research follows available data, not real clinical needs.
Looking forward to being back in Berkeley and seeing you all!
Thanks for hosting me
"Cementing a machine-learning ecosystem: scikit-learn and beyond": on Friday March 20, we at @ucbids.bsky.social and the Berkeley Statistics department are delighted to host @gaelvaroquaux.bsky.social for a seminar. Join us in person or online!
events.berkeley.edu/BIDS/event/3...
La pétition zrr.collectif-inria.fr a dépassé les 900 signataires. On commence à faire des stats, elle a en particulier été signée par 40% des chercheur·euses permanent·es rémunéré·es par l'Inria (300 personnes) et 33% des responsables d'équipes projets.
The next scikit-learn release will allow inspecting the type and values of attributes of fitted estimators in Jupyter notebooks & example code rendered as HTML in sphinx-gallery powered project websites.
scikit-learn.org/dev/auto_exa...
