Comment on Cleaning algorithm finds 20% of errors in major image recognition datasetsparentComments−strbean6yAlso, this is applies to mislabeled data in your training set, right? Not a good thing if it is in your test set.
Comments
Also, this is applies to mislabeled data in your training set, right? Not a good thing if it is in your test set.