I am training an AlexNet Convolutional Neural Network to classify images in a dataset. I want to know if there is any general rule for using data augmentation in training a neural network. How can I make sure about the amount of data, and how can I know if I need more data?

  • $\begingroup$ Please, put your specific question in the title. $\endgroup$ – nbro Dec 15 '20 at 18:21
  • $\begingroup$ @nbro edited, thanks. $\endgroup$ – SahaTib Dec 15 '20 at 18:28
  • $\begingroup$ It's not very clear the question in the title though. What do you mean "rule for number of applying"? Maybe your question is: "Is there any rule of thumb to determine the amount of data needed to train a CNN? Moreover, it seems that you're asking another question: "Which data augmentation techiniques should you use? Is there any rule?". Please, if you have two question, ask each of them in its separate post. $\endgroup$ – nbro Dec 15 '20 at 18:31
  • $\begingroup$ @nbro No, I just want to know about the amount of data, I edited the title again, sorry for my english. $\endgroup$ – SahaTib Dec 15 '20 at 18:43
  • 1
    $\begingroup$ Your question seems quite broad and unclear – especially the added question about data augmentation. My understanding is that there is no "rule of thumb"; the amount of data you need depends on the complexity of the images that you are / will be dealing with. Here are some sources on data augmentation: journalofbigdata.springeropen.com/articles/10.1186/… , arxiv.org/abs/1712.04621 , arxiv.org/abs/1609.08764 $\endgroup$ – The Pointer Dec 16 '20 at 9:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.