Frequently Asked Questions
Is MIDST an individual or team competitions?
You are welcome to participate individually or as a team. We don’t have a cap on the team size.
What is the distinction between the train and holdout dataset?
The training set is a subset of a bigger (public) population data and the holdout data set is the remainder of that population data. They are not necessarily the same size. The challenge/target data points are an even mix of members (selected from the training set), and non-members (selected from holdout). So if we have 200 challenge points, 100 are chosen from the training set and 100 are chosen from the holdout set.
Is the population / auxiliary data available in the black box and/or white box tasks?
Yes, the population dataset is public, for both black-box and white-box tasks.