Re:A popular self-driving car dataset is missing labels for ...
by jwandborg on Thursday February 13, 2020 @02:51PM
(#22320835)
It's hard to reach 0% bad labels because:
1. You can't have an infinite amount of good labels
2. Humans are in charge of labeling too.
The question is if you can reliably overcome the number of bad labels in your training set, so that 33% of bad labels equates to <33% "insanity" in the system.
1. You can't have an infinite amount of good labels 2. Humans are in charge of labeling too.
The question is if you can reliably overcome the number of bad labels in your training set, so that 33% of bad labels equates to <33% "insanity" in the system.