I want to create a small dataset (about 10 classes and 20-30 images each), should I add some noise (wrong label sample) in the training, validation and test datasets, and why?
Asked
Active
Viewed 31 times
1
-
Hi. Are you asking whether you should or not deliberately add some noise to the dataset? I've added the word "add" to your question, but I am unsure if this was the actual word missing. – nbro Dec 26 '19 at 13:10
-
1What is your purpose in creating the dataset? What do you hope to achieve? We need to know in order to answer whether adding noise would be useful – Neil Slater Dec 26 '19 at 13:41