Workshop See4C Spatio-temporal Series -Machine Learning Aix-Marseille hors série

This is a past event

21 people went

Details

Workshop and Hackathon spatio-temporal time series forecasting

Invited speakers: Mehdi Mirza (Google DeepMind), Bill Lotter (Harvard, Univ., USA), Danny Silver (Acadia University, Canada), Benjamin Donnot (RTE: Réseau de transport d’électricité, France), Sergio Escalera (U. Barcelona, Spain)

AND

Challenge in Video-Forecasting: replacing missing frames in tele-conferences.

Following the success of the Paris ML Meetup Hors série #6 Season 4: See.4C Spatio-temporal Series Hackathon (https://www.meetup.com/fr-FR/Paris-Machine-learning-applications-group/events/237019334/?eventId=237019334), the See.4C project of spatio-temporal forecasting organizes a second round of the “warm up challenge” whose objective is to test a new version of the platform of a large-scale upcoming EU challenge to predict electricity flows in the French power network (2 millions Euros in prizes). This is an opportunity to get a head start and learn about spatio-temporal forecasting and practice using GPUs!

The task of the “warm up challenge” is of real practical interest: predicting upcoming frames in video data. The applications may include replacing missing frames in a tele-conference when data transmission is defective.

For this benchmark, we are using thousands of videos of speakers facing a camera, sampled at 25 frames per seconds. We limit the resolution to small 32x32 pixels frames in black and white to permit obtaining results in a short time, in the context of a hackathon.

The problems far from trivial. The baseline method consisting of “freezing” the person and making constant prediction is very difficult to beat. While computer vision methods may help, it will be interesting to see whether generic methods applicable to other domains can significantly outperform the baseline, including deep learning methods. Classical signal processing methods such as ARIMA models should also be serious contenders.

New in this edition: we will provide the participant with a starting kit that includes deep-learning methods and teach them to use GPUs.

More information on the See.4C project website (http://see4c.eu/workshop/4/description/)

A doodle to organize the coming on saturday morning (carshare, bus, train) has been setup: http://doodle.com/poll/errxkpbg93zzc7bf

Contact information

[masked]

Organizers:

Scientific board
Florin Popescu
Sergio Escalera
Xavier Baró
Stephane Ayache
Cecile Capponi
Isabelle Guyon

Co-organizers and coaches
Julio Jacques Jr.
Lisheng Sun
Diviyan Kalainathan
Sébastien Treguer

Local organizers:

Cecile Capponi
Olivier Guillaume
Laurent Cetinsoy