New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Divergence between paper scaled sigmoid and package scaled sigmoid #404

Open

r-dion opened this issue Dec 7, 2023 · 0 comments

r-dion commented Dec 7, 2023 •

edited

Loading

Hello,

Firstly, thank you for your work !

I am using your NAB metric, and exploring its implementation I found some divergences around the scaled sigmoid.

In the NAB paper (Lavin2015), the scaled sigmoid is defined as

for Atp in [0,1] and Afp in [-1,0].

In the implementation, the scaled sigmoid is defined as follows (skiping the docstrings):

def scaledSigmoid(relativePositionInWindow):
  if relativePositionInWindow > 3.0:
    # FP well behind window
    val = -1.0
  else:
    val = 2*sigmoid(-5*relativePositionInWindow) - 1.0

  return val

Is it deliberate that (Atp - Afp) is replaced by a constant (2 is this case) ? It changes some results.

Thank you for your time.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment