Pour vous authentifier, privilégiez eduGAIN / To authenticate, prefer eduGAINeu

RAMP 09 Pollinating insects

La Paillasse

La Paillasse

226 Rue St Denis, 75002 Paris
Akin Kazakci (MINES ParisTech), Balázs Kégl (LAL), mehdi cherti (LAL/Appstat)

The Paris Saclay Center for Data Science is organizing a RAMP on image classification and deep learning. We will classify images of pollinating insects from the SPIPOLL crowdsourcing project of the Paris Museum of Natural History (MNHN).

The kick-off event will be on May 29 at La Paillasse. The RAMP will run in competitive mode (scores available on the leaderboard but not the submission codes) until 20h June 7. We will switch to the open collaborative mode starting June 8 at the event organized by Proto204 in the Protobus at Futur en Seine. The prizes of the top three contestants of the the closed phase (three pots of honey, brought to you by pollinating insects) will be awarded at the June 8 event.

The kick-off event will start with a half-day tutorial on deep learning and convolutional neural nets, followed by an introduction to the particular problem to solve, the RAMP platform, and the starting kit. The tutorial is intended to an audience with basic notions of the pydata ecosystem and machine learning. The tutorial will be based on the courses of Andrey Karpathy and Olivier Grisel/Charles Ollion.

La Paillasse has a limit of 40 participants. To enter to the kick-off/tutorial of May 29, we will ask you to sign up at the RAMP platform, sign up to the Titanic test RAMP, and make a submission. The selection among those who clear this step will be based on first-come-first-serve, but we maintain the right to change the criteria. If there is enough interest, we can add a second day later in the week.

Note that participation in the pollinating insects RAMP is not conditioned on participating in the kickoff event. The starting kit is entirely self-explanatory. 

We have not yet set up the pollinating insects event at the RAMP site, but those who would like to start preparing can download the starting kit for our previous insect classification RAMP (which had less images and classes). The starting kit contains a Jupyter notebook which you can open on your laptop. On the other hand, to execute the starting kit, you will need to have access to a GPU-equipped machine. The simplest way is to set up an account at AWS following this tutorial. We have prepared an AMI called "pollinating_insects_users" on the Frankfurt and Oregon nodes which contains the data and the starting kit preinstalled.

You can read more about RAMPs here.

Consider joining our slack team. You can receive information about ongoing and upcoming RAMPs there, share information with us and fellow rampers during and after RAMPs. You can also get help from and give advice to others. Finally, you can give us valuable
feedback to improve the platform. 

We are grateful to the Université de Champagne-Ardenne ROMEO HPC Center and Amazon Web Services for providing the GPU backend and engineering support for the RAMP, to La Paillasse for hosting the event, and to NVIDIA for sponsoring the event.

Below you can sign up to the launching event on May 29. Don't forget to also sign up at the RAMP platform.

  • Aishik Ghosh
  • Alexandre Boucaud
  • Amine Tamasna
  • Anaelle Laurans
  • Andry Ramboalimanana
  • Arnaud Lecoules
  • Ashley Hill
  • Ayoub Ghriss
  • Balázs Kégl
  • Benoit Hubert
  • Boris Muzellec
  • Diego DE SOUZA
  • dorra ELABED
  • Erwann Martin
  • gunavadh lim
  • Inès BAHRAM
  • Issam Benabid
  • Jonathan Lumbroso
  • Joris Van den Bossche
  • Julien Guillaumin
  • karim Ould Aklouche
  • Men HUANG
  • Moha Hillali
  • Mounya BELGHITI
  • Nicolas SALEILLE
  • Nikhil Shetty
  • Okay Gunes
  • Omar Mohammed
  • Roman Yurchak
  • Rudy Delouya
  • Rémi Flicoteaux
  • Sebastien Treguer
  • Sefiane Touami
  • Simon FRANCOIS
  • Soobash Daiboo
  • Théophile Sanchez
  • Victor Estrade
  • Virginia Azzolini
  • wandrille guesdon
  • Yaohui WANG
  • Yetkin Yilmaz
  • youssef maghzaz
    • Welcome of participants & Coffee
    • Deep learning and convolutional neural nets tutorial
      • 1
        Setting up AWS
        Speakers: Balázs Kégl (LAL), Mr mehdi cherti (LAL/Appstat)
      • 2
        Convnet tutorial
        Q&A on Andrey Karpathy's course: http://cs231n.github.io/convolutional-networks/
        Speakers: Balázs Kégl (LAL), Mr mehdi cherti (LAL/Appstat)
      • 3
        The pollinating insect starting kit
        Speakers: Balázs Kégl (LAL), Mr mehdi cherti (LAL/Appstat)
      • 4
        Speaker: Guillaume barat (NVIDIA)
    • Buffet lunch
    • Pollenating insects and the RAMP platform: First submissions
      • 5
        The RAMP: motivations and goals
        Managing the data science process.
        Speaker: Dr Akin Kazakci (MINES ParisTech)
      • 6
        The RAMP: where we are going
        Speaker: Balázs Kégl (LAL)