Is there a target number of classifications per image you are shooting for?
-
by heracleum
From the blog post it seemed that 600,000 classifications was your target for each data set - but I was curious what that worked out to per image. And whether you were trying to develop a "confidence factor" for the classifications of the same image based on the degree to which the classifications agreed or disagreed.
Posted
-
by ttfnrob admin, scientist
Hi - there isn't a fixed number of classifications per image. If more than 5 of the first ten people say that the image is blank - we retire it. If not then we ask a total of 50 classifications. We then look at the data and see if we need t push that higher or not. This lets us concentrate on getting more classifications from volunteers for the interesting images. I hope that helps explain 😃
Posted
-
by SundogWE
If you THINK you have found a galaxy and mark it as such and it turns out to be one, how will you ever know? Is there a feedback system if you find valuable classifications in the images?
Posted
-
by heracleum
Thanks - I was wondering whether you "retired" images -- during the first round of images, my subjective impression was that they appeared to get more "interesting" as time went by and this confirms the reason!
Posted