-1

I am doing video classification with a model combining CNN and LSTM. In the training data, the accuracy rate is 100%, but the accuracy rate of the test data is not so good. The number of training data is small, about 50 per class. In such a case, can I declare that over learning is occurring? Or is there another cause?

pie
  • 155
  • 4

1 Answers1

1

To me, accuracy on a training set makes no sense. I ignore it. If you want to test overfitting on training data, you can do this by cross-validation. For details, please see the link

Waqas
  • 386
  • 3
  • 15
  • When cross validation is carried out, what kind of accuracy rate value is given when the model is overfitting? – pie Jul 21 '18 at 09:02
  • Please keep in mind that overfitting is not just a yes/not condition. Very different test and train accuracies mean that overfitting is serious. – Pere Jul 21 '18 at 10:08