Closed tanaka-jp closed 4 years ago
Hello,
Thanks for your interest in the paper! :)
The confusion matrix is actually computed on the validation data, which is a small proportion of the training data that we set aside. See this line in the compute_confusion_matrix function.
Thanks for answering.
Hello. I like your paper in which the new idea of curriculum learning was proposed.
But I must ask you a question.
Which did you get the confusion matrix from, training data or test data? I think that getting the confusion matrix from test data is a kind of data leakage.