It seems to be a take on ROC and could be phrased more humorously. Replacing either acceptance/denial with "filter" makes it not at all insightful,so it is at least a step up from tautology.
In the acceptance case, author presumes(?) that the designer looks only at false positives, etc. there's deep stuff here if people want to dig (postselection fallacies happen in the classical world too!) but poster might have been high on something.
I don't see a world* in which allowing the irreversible flagging of this is positive sum :)
*I only see one in which HN is training some model on the commentariat but not on the "fast-grants" (that's my label for it rn) based policy optimization itself :)
To the posted paper. top schools asking interesting questions, great, if I were to try to make a useful prediction it is that this is going into a second tier conf at best, this seems to be "owned" by junior members of the pjlab team (disregard the corresponding author), is a clear intern project, just a tier below "fast-grants" eligibility.
Chinese schools have weird policy optimization tricks :)
Should we ever know what kind of person flagged this
https://news.ycombinator.com/item?id=45580780
:)
It seems to be a take on ROC and could be phrased more humorously. Replacing either acceptance/denial with "filter" makes it not at all insightful,so it is at least a step up from tautology.
In the acceptance case, author presumes(?) that the designer looks only at false positives, etc. there's deep stuff here if people want to dig (postselection fallacies happen in the classical world too!) but poster might have been high on something.
I don't see a world* in which allowing the irreversible flagging of this is positive sum :)
*I only see one in which HN is training some model on the commentariat but not on the "fast-grants" (that's my label for it rn) based policy optimization itself :)
To the posted paper. top schools asking interesting questions, great, if I were to try to make a useful prediction it is that this is going into a second tier conf at best, this seems to be "owned" by junior members of the pjlab team (disregard the corresponding author), is a clear intern project, just a tier below "fast-grants" eligibility.
Chinese schools have weird policy optimization tricks :)