Time Series Forecasting Grand Competition

A workshop on the 1st NN GC Forecasting Competition will be held at the 2009 International Joint Conference on Neural Networks (IJCNN'09) June 14-19, 2008, Atlanta, Georgia, USA. Abstracts are invited.

Submissions to the new 2009 NN GC1 competition

A set of 6 small datasets of transpotation data are available at http://www.neural-forecasting-competition.com/. The competition data consist of different datasets, each with 19 time series of a different time series frequency of yearly, quarterly, monthly, weekly, daily, and hourly data. A competitor must submitt at least predcitions on one dataset (reduced set of 19 series) of the 6 datasets to be eligible to participate in the workshop. Over the course of the competition competitors will be encouraged to submitt predictions on all 6 datasets (complete set of 114 series). All those planning to submit predictions on at least one datasets or all 6 of the datasets are invited to submit abstracts describing their methods for presentation at the 2009 IJCNN conference.

Submissions to the past NN3 & NN5 competitions

In addition we also invite participants from the past NN3 and NN4 competitions to submitt abstracts and present their earlier submissions at IJCNN'09. Our aim is to provide those that have participated in previous competitions at the IJCNN'07 and WCCI'08 or ISF'07/08 and DMIN'07/08 conferences, and that were unable to attend before, with the opportunity to present their original methodology and possible extensions and revisions since then . We hope this will disseminate best practices on how to predict time series data of different time frequencies to a wider audience, in particular due to its relevance for the corresponding datasets with similar time series frequency in the 1st NN Grand Challenge (NN GC1).  Submissions are limited to 1 abstract per competition, so a maximum of 3 presentations.

Abstract Submission 

Abstracts are limited to 750 words in any format as MS Word document or as PDF, and should describe the method and methodology used to predict the time series . The title should indicate the method used and contain the name of the competition (e.g. 'NN3', ‘NN5’ or 'NN GC1' in the title, submissions as plain text via email are encouraged (submitt to mhtml:{995876DF-F803-4C1F-82AB-E0BC3D7A66C7}mid://00000096/!x-usc:mailto:competition-chair@neural-forecasting-competition.com). Abstracts will be peer-reviewed for relevance, orginality, significance, clarity and technical quality. Submisison deadline is Friday 29th May 2009. Short extensions may be granted.

Workshop Presentations 

–          Presentations are restricted to NN3, NN5 and NN GC1 competition participants (those that have submitted predictions on the NN3 or NN5 datasets in the past or will submitt predictions for the NN GC1 by the deadline). Presentations of those submitting abstracts without predictions at NN3, NN5 or NN GC1 (by the prediction submission deadline of Monday 5th June 2009) will be cancelled and will not be eligible to present.

–     Presentations held in the past at other NN3, NN5 or NN GC conference sessions (e.g. ISF'08 and DMIN'08) do not preclude presentation at IJCNN'09 – you can present at multiple conference sessions, and we welcome your input to the IJCNN community!

–          In order to present at the NN GC1 session those accepted must register IMMEDIATELY to attend the workkshop following the IJCNN'09 conference at the cost of US$ 70.-  Those already registered for IJCNN'09 must register for the workshop seperately. Those not registered for IJCNN'09 do not need to register for the main conference. Please register immediately to ensure your place at the workshop.

Important Dates

29 May 2009:    Abstracts due (extended)
31 May 2009:    Notification of acceptance (extended) 
1   June 2009:    IJCNN'09 final registration deadline
5   June 2009:    Predictions due (extended)

If you are interrested, but are running into problems with your submission or the deadlines, please contact the organisers immediately.

Read more at http://www.neural-forecasting-competition.com/