Is it correctly understood that in "Large Scale Visual Recognition Challenge" the competitor's biweekly runs ahead of the final submission is against the set that actually determines the final result of the contest?!
That's what I was wondering. If so, submitting many test runs makes it trivial to provide a submission that works perfectly against the test data specifically.
...for some definitions of "trivial" and "perfect". At this level, I suspect even a small advantage would result in winning the contest, which is the point here.
