#trackml search results

... with experience from #higgsml and #trackml competitions ! #hepml

It is much more a foundation library with very efficient C++ code for Runge Kutta integration for 3D track extrapolation in non uniform magnetic field, track fitting with Kalman filtering. Plus track simulation capabilities. #trackml was about track finding which is next

... and it’s out! We finally manage to release the final « Throughput » paper on #trackml challenge. Check it out to see how participants achieved one order of magnitude speed up through amazing tricks. arxiv.org/abs/2105.01160 #hepml

My office door at @ijclab @UnivParisSaclay I can’t wait to do real workshops again, so than I can redecorate. (But not only). #academictwitter #hepml #trackml

I concur, for our #TrackML #kaggle challenge we've asked all competitors to submit documented code. A dozen did. We got a basket loads of new (to our community) ideas from the top 20. Top 1 was unpractical. We even noticed that #100 was doing very well for a tiny sub-problem.

And in order to celebrate the conclusion of #trackml and the excellent organisation of @ichep2020 I even made a custom virtual background!

.@SaltyBurger gives tomorrow the conclusion talk on #trackml challenge (#kaggle and #codalab ) at #ichep2020 the major (virtual) physics conference ichep2020.org . One more paper to release and now many spin offs...

Several talks using #trackml #kaggle challenge, in particular by Kiehn Basara Tuysuz Choma.

#trackml challenge talk accepted at @ichep2020 ! That’s the final final one ! Now conf will either be remote end July or postponed to Jan/Feb 2021. #kaggle #codalab.

Look ! Our #trackml #kaggle challenge paper in #NeurIPS2019 (no 2018!) Competition book, for real ! @SaltyBurger @vlimant @dhpmrou @AMROUCHEsabrina #hepml

Vancouver #neurips2019 ! Will be presenting the final conclusion of our #trackml #kaggle and #codalab challenges at CiML workshop Friday.

.@SaltyBurger trying out the #trackml visualisation on #hololens. Trackml bunch of tracks is hovering over the audience at institut pascal. Andreas even walked into it

This is the workshop series where #trackml was initiated

The connection is finding thin 1D objects in a 3D tensor #trackml #hepml

Fascinating talk from Hugues Talbot prof @centralesupelec at our tracking workshop at Institut Pascal Paris Saclay indico.cern.ch/event/847626/t… . We realized there could be a connexion between finding blood vessels in the brain and tracks in a Liquid Argon TPC #hepml #trackml

There will be in depth discussions about #trackml #kaggle and #codalab challenges lessons and much more. For example possible use of @LightOnIO OPU or quantum computer for tracking.

Happy to annouce that my @trackmllhc poster won the MIT Press prize at @DeepIndaba ! #sautiyetu #DLIndaba2019 #Trackml #HEP

Slides of our #TrackML grand finale workshop at @cern : indico.cern.ch/event/813759/. Special thanks to the three best performers of Throughput phase on #codalab, and some of the best performers of accuracy phase on #kaggle for their talks. Hope you enjoy the @ATLASexperiment visit !

Not one, but two #trackml talks at the major #epshep2019 physics conference eps-hep2019.eu . David Rousseau @dhpmrou in outreach session and Laurent Basara in data analysis session. Friday morning. Don’t miss us! #hepml #kaggle #codalab

#trackml #hepml #kaggle #codalab @cern Slides on : indico.cern.ch/event/813759/

Day 18 of #100DaysOfMLCode. My model grows stronger! #kaggle #trackml #python

Day 24 of #100DaysofMLCode currently processing a submission for #kaggle #trackml on a 72 core (vcpu) #aws box 📦 #parallelismftw #python #Machine_Learning #DataScience

Ce Lundi ! Paris Machine Learning meetup Hors série. Le #trackML challenge bootcamp meetu.ps/e/Fm5wF/1k82H/d

3/8 Same picture from number of submission per day #trackml

4/8 ... but not so many last minutes new teams: #trackml

Day 14 of #100DaysOfMLCode experimenting with different features and model architectures for the #kaggle #trackml competition and wishing there was an a #python implementation of DBSCAN that could run on my #gtx1080 #GPU

Day 23 of #100DaysofMLCode spent the last few days parameter tuning and working on ways to improve validation and submission speed using parallelization for the #Kaggle #trackml competition.

Day 33 of #100DaysOfMLCode The results are in! I landed at 50th place on the private LB of #TrackML comp getting a silver medal and my first medal on #Kaggle. Really enjoyed this competition and learn a lot!

Day 32 of #100DaysOfMLCode New submission in. Managed to gain back a few positions on the leaderboard. At 43rd now but still a day left and results on the private LB could be different! #kaggle #trackML

Day 6 of #100DaysOfMLCode more EDA and digging through the #kaggle forums for the #trackml competition. Who knew particle detectors could be so complicated!

Can #MachineLearning assist high-energy physics in discovering and characterising new particles at the #HiLumiLHC? Participate in the #TrackML challenge if you think you have the answer: cern.ch/go/ZH7Z

Happy to annouce that my @trackmllhc poster won the MIT Press prize at @DeepIndaba ! #sautiyetu #DLIndaba2019 #Trackml #HEP

Last #trackml connect-the-dots exercize for #CHEP2018 Can you find a better track, @dhpmrou ?

Getting Challenged to succeed #TrackML #CERN seminar

Prof. Cecile Germain talks to us about the #TrackML challenge and the reason why we should help High Energy Physics with Machine Learning

@dhpmrou gives the summary report of #trackml at the @CERN IML workshop @trackmllhc @kaggle @Common6round @nvidia @CERNopenlab @iris_hep @PyTorch

So how it works: you're given 3D points coordinates, which you have to group into tracks which are arc of helices : 100.000 points into 10.000 tracks.Very simplified 2D view #trackml

Look ! Our #trackml #kaggle challenge paper in #NeurIPS2019 (no 2018!) Competition book, for real ! @SaltyBurger @vlimant @dhpmrou @AMROUCHEsabrina #hepml

Over .90 on the leaderboard for the #TrackML @kaggle challenge. 2 days left to form your team. 9 days left to push a solution. go Go GO #MachineLearning !

NVIDIA booth at #SC18 running DBSCAN and visualization of the #trackml dataset.