Ants as tools to understand society
By Penn State University
about 2 years ago
In the overview is said the following: The training data provides the coordinates of a subset of the ants for all the time frames, and the goal of the challenge is to predict the coordinates of the rest of the ants for all the time frames.
But at least in my train dataset I have the positions of all the ants for the first half of the frames, and in the test set the positions for all the ants in the second half of the frames. Is that right?
about 2 years ago |
You are right. The training dataset is indeed the position of all the ants for the first half of the frames, and the submission is about predicting the positions of all the ants for the rest of the frames. We have updated the challenge description to avoid any further confusion.
In any case, the evaluation function is agnostic to the way the dataset is divided (because it maps the predicted coordinates of the ant-id and frame-id pairs to that of the answer file), and hence, the already submitted submissions are graded as they should be.
Thanks for pointing it out though,