DeepSense-ITU Multi-Modal Beam Prediction Challenge

Test and Adaptation Sets

Competition Test Set Download Links

The structure of the test set is the same as the development dataset. You can find more information about the structure here.

Adaptation Dataset

To further facilitate the generalization and adaptation task in this competition, we provide a small adaptation dataset consisting of the following:

  • 50 labeled samples from the unseen Scenario 31
  • 25 labeled samples from Scenario 32
  • 25 labeled samples from Scenario 33

The sequences in this adaptation dataset have the same sampling rate as the competition test set. 

The purpose of the adaptation dataset is to provide a very small set of data points from the unseen distribution (that exists in the test set). It is meant to augment the existing development dataset and not to replace it. It is up to the participants to use or not use this dataset. Further, the participants have the flexibility to utilize this sample set in the way they decide on, and this is part of the ML model design/training process.  

For example, a team may choose to refine the previously trained model with this adaptation set, another team may choose to mix the training and adaptation sets before training the model, and one other team may make a different decision on how to utilize this small set of samples. 

If you have any questions, please contact the organizers using competition@ or post your question on the DeepSense Forum