Fullscreen
Loading...
 
Tao
Print

Seminar06022018

March, Tuesday 13th (date changed because of snow storm)

14:30 (room 2014, 'Digiteo Shannon' 660 building) (see location )

David Rousseau

(Laboratoire de l'Accélérateur Linéaire (LAL), Orsay)

Title: TrackML : The High Energy Physics Tracking Challenge


Abstract

We organize on the Kaggle platform a data science competition to stimulate both the ML and HEP communities to renew core event reconstruction algorithms in preparation of the next generation of particle detectors in the Large Hadron Collider at CERN.
With event rates already reaching hundred of millions of collisions per second, physicists must sift through ten of PetaBytes? of data per year.
The outcome of this challenge will likely allow the LHC to fulfill its rich physics program, understanding the characteristics of the Higgs boson, searching for the elusive dark matter, or elucidating the dominance of matter over anti-matter in the observable Universe.

In a nutshell : one event is like an image with 100.000 3D points ; how to associate the points onto 10.000 unknown approximately helicoidal trajectories ? avoiding combinatorial explosion ? you have 10 seconds. But we do give you 100.000 events (>100GB) to train on.

HEP classical algorithms to solve this problem include Kalman filters and Hough transforms, while a number of CS techniques can be invoked: MCTS, LSTM, clustering, CNN, geometric deep learning and more.
This truly large scale challenge will be conducted in two phases focusing first on accuracy (starting in March 2018) and secondly on throughput (starting in the summer).



Contact: guillaume.charpiat at inria.fr
All TAU seminars: here


Contributors to this page: guillaume .
Page last modified on Monday 05 of February, 2018 16:34:59 CET by guillaume.