A Method of Data Label Checking and the Wrong Labels in MNIST and CIFAR10
4 Pages Posted: 20 Nov 2017
Date Written: November 16, 2017
Abstract
The quality of data label is an important issue of machine learning. Data label checking is done by manual, which is not efficient and may still have wrong labels. I introduce a method to check label quality and 3 wrong labeled in 2000 digits pictures of MNIST, 6 wrong labeled in 2800 pictures of CIFAR10 and find out 64 wrong labeled in 400 digits pictures of NIST SD19v2.
Suggested Citation: Suggested Citation
Zhang, Xinbin, A Method of Data Label Checking and the Wrong Labels in MNIST and CIFAR10 (November 16, 2017). Available at SSRN: https://ssrn.com/abstract=3072167 or http://dx.doi.org/10.2139/ssrn.3072167
Do you have negative results from your research you’d like to share?
Feedback
Feedback to SSRN
If you need immediate assistance, call 877-SSRNHelp (877 777 6435) in the United States, or +1 212 448 2500 outside of the United States, 8:30AM to 6:00PM U.S. Eastern, Monday - Friday.