search

mmasana / FACIL

Framework for Analysis of Class-Incremental Learning with 12 state-of-the-art methods and 3 baselines.
https://arxiv.org/pdf/2010.15277.pdf
MIT License
512 stars 98 forks source link

accuracy #35

Open tb626 opened 1 year ago

tb626 commented 1 year ago

Does the accuracy of each stage only use the category validation set of the current task? Should the accuracy of all stages of training be taken as the final result? Thank you very much for your reply

mmasana commented 1 year ago

I'm not sure if I understand the question, but the accuracy when learning each task is only calculated from the samples of classes belonging to that task. During training, the accuracies are from train or validation. During the final evaluation, those are calculated from test for benchmarking. Then, the average of all accuracies of all tasks should be taken as the final result. Hope that helps!