GMvandeVen/brain-inspired-replay

Precision or Accuracy?

kaustubholpadkar opened this issue · 4 comments

kaustubholpadkar commented

When I run the experiments, it displays as follows:

| Task: 3/5 | training loss: 0.407 | training precision: 0.992 |: 100%|███████████████████████████████████| 2000/2000 [08:37<00:00, 3.87it/s]

On-the-fly visdom plots also show precision as y-axis but the title of the plot is average accuracy.

Can you please clarify on that? I have seen same confusion in the code of your other repo: https://github.com/GMvandeVen/continual-learning

GMvandeVen commented

Hi, thanks for your question. In this code (and also in the code of https://github.com/GMvandeVen/continual-learning) I have been using the terms "accuracy" and "precision" interchangeably; with both I mean the same thing. Sorry for the confusion! When I have some time I will try to make the usage of these terms in the code more consistent.

kaustubholpadkar commented

Thanks for quick response. I understand that you have used the terms "accuracy" and "precision" interchangeably. But both have different theoritical meaning.

  • Accuracy = (TP+TN)/(TP+FP+FN+TN)
  • Precision = TP/(TP+FP)

I just want to confirm which one of accuracy or precision is used in these papers.

GMvandeVen commented

Thanks for pointing that out, that is very helpful. Sorry, I realise I should have been more careful with these terms.
The performance measure that is reported in this paper (and also in my other ones) is the proportion of samples classified correctly.
I just updated the code and I now only refer to this measure as accuracy. (I'll update the other repositories as well.)

kaustubholpadkar commented

Thank you so much!