Predicting Generalization in Deep Learning Forum

Go back to competition Back to thread list Post in this thread

> About the dataset in the tasks

Is the data-set in task1_v4 and task2_v1 suitably far apart from each other as these both are from task4 and task5?
Suppose there be a distance metric dist(d1, d2) which gives a hypothetical distance between datasets, can we assume dist(d_i, d_j) to be almost similar for d_i, d_j \in {task1_v4, task2_v1, task4, task5} ?

Posted by: shmakn @ July 22, 2020, 4:35 a.m.

That's a deep questions to which I do no have a good answer. After all, it's not clear to me what kind of metric it would even look like.
I will say, however, that they do not share supports.

Posted by: ydjiang @ July 22, 2020, 4:44 a.m.
Post in this thread