Machine Learning master project¶
Objectives¶
- Foster collaboration between IN2P3 labs and experiments on ML activities
- Share state-of-the-art knowledge between project members and spread it across IN2P3
- Encourage physicist to participate to ML conferences
- Document ongoing ML projects and HEP related ML tutorials
Coordination¶
- David Rousseau, IJCLab, Orsay – scientific lead
- Thomas Vuillaume, LAPP, Annecy – technical lead
- Julien Donini, LPC, Clermont – AI scientific delegate for IN2P3 (since 2024)
News¶
-
Julien gave a panorama of IN2P3 ML activities at the Journées du CCIN2P3 in June 2024.
The slides can be viewed here. -
AISSAI Workshop on Heterogeneous data end of September
The last IN2P3 AISSAI workshop was organised by the L2IT, Toulouse between Sept 30 and Oct 3rd 2024.
The presentations can be found online. -
IN2P3/IRFU ML annual workshop Our annual workshop between IN2P3 and CEA/IRFU was held on November 20-22 2024.
We had the pleasure to be hosted by Giulio, Jérôme and Eric at IPHC in Strasbourg.
The slides and records can be found in the indico page here.
Getting started¶
Here is a list of the point of contact in each lab regarding the activities of this project.
Alexandre Boucaud < alexandre [dot] boucaud [at] apc [dot] in2p3 [dot] fr >
Freddy Poirier < poirier [at] arronax-nantes [dot] fr>
Sébastien Gadrat < sebastien [dot] gadrat [at] cc [dot] in2p3 [dot] fr >
Julien Zoubian < zoubian [at] cppm [dot] in2p3 [dot] fr >
Adnan Ghribi < adnan [dot] ghribi [at] ganil [dot] fr >
David Rousseau < rousseau [at] ijclab [dot] in2p3 [dot] fr >
Jérôme Pansanel < jerome [dot] pansanel [at] iphc [dot] cnrs [dot] fr >
Guillaume Baulieu < g [dot] baulieu [at] ipnl [dot] in2p3 [dot] fr >
Sylvain Caillou < sylvain [dot] caillou [at] l2it [dot] in2p3 [dot] fr >
Thomas Vuillaume < thomas [dot] vuillaume [at] lapp [dot] in2p3 [dot] fr >
Frédéric Magniette < frederic [dot] magniette [at] llr [dot] in2p3 [dot] fr >
Frédéric Druillole < frederic [dot] druillole [at] lp2ib [dot] in2p3 [dot] fr >
Julien Donini < donini [at] cern [dot] ch >
No contact yet!
Anja Butter < butter [at] lpnhe [dot] in2p3 [dot] fr >
Pierre-Antoine Delsart < delsart [at] in2p3 [dot] fr >
Johann Cohen-Tanugi < johann [dot] cohen-tanugi [at] in2p3 [dot] fr >
A good starting point would be to contact them to let them know what you are working on or plan to be working on.
If you are getting started with machine learning in general, we recommend to have a look at the Fidle course series (in French) or at the content of the latest schools organized by IN2P3.
Mailing list¶
Used for asking specific questions or sharing information about upcoming events.
How to contribute?¶
This website has been designed to be built collaboratively.
We need your help to maintain up-to-date sections such as publications, projects and events.
There are 3 ways you can contribute:
1. Adding new content¶
-
To contribute new projects, please fill the form at this link.
It will automatically add your project to the relevant page during the next website build (every Sunday). -
To contribute new PhD theses, please fill the form at this link.
It will automatically add the thesis to the relevant page during the next website build (every Sunday). -
To contribute a new publication, please fill the form at this link.
We will manually add it to the bibtex file in the repository.
2. Reporting an issue or requesting a modification in the website¶
To report an issue or propose a modification in the pages, you may open an issue with all the needed information to be added to the website (check the current pages to know what you need to provide).
To do so, you will need to connect to the GitLab platform using your eduGAIN credentials.
3. Propose changes yourself¶
If you want to propose important changes to the pages, please start a discussion first by opening an issue and tag @vuillaume.
Smaller changes can be proposed directly through merge requests:
- Request to join the project on GitLab (up top, near the project ID).
- Create a new branch
- Modify the markdown files in the
/docs
directory or thein2p3ml_publications.bib
- Test locally (see below)
- Open a merge request and assign it to
@vuillaume
If you include new markdown pages, make sure they are listed in the
nav
section of the [mkdocs.yml
] file in the root directory.
To develop and test the webiste, please follow the instructions below.
Website development¶
The static website is build with Python and mkdocs
.
The list of publications is build with bibtex2html
and included in the website a posteriori using a Python script.
Local setup¶
- Create a virtual environment (optional but recommended)
python -m venv venv source venv/bin/activate
- Install the dependencies
python -m pip install -r requirements.txt
- Install
bibtex2html
(instructions here)
Local testing¶
- Build website with live-reloading server for development
mkdocs serve
- Strict building of the website (check)
make test-site
- Build publication list (check)
make test-biblio
- Build website with publication list
make build
- Change the bibliography style
make build bibstyle=<mystyle> # e.g. to use a <mystyle>.bst file in the current directory