MycroftAI
/
mycroft-precise
mirror of https://github.com/MycroftAI/mycroft-precise.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

Fix Nan in false_neg and false_pos metrics 

This was caused when small batches with no positive samples caused a divide-by-zero
pull/19/head
Matthew Scholefield 2018-07-11 19:34:22 -05:00
parent d8dd0f3240
commit 36804075b3
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
precise/functions.py
View File

@ -51,12 +51,12 @@ def weighted_mse_loss(yt, yp) -> Any:
def false_pos(yt, yp) -> Any:
from keras import backend as K
return K.sum(K.cast(yp * (1 - yt) > 0.5, 'float')) / K.sum(1 - yt)
return K.sum(K.cast(yp * (1 - yt) > 0.5, 'float')) / K.maximum(1.0, K.sum(1 - yt))
def false_neg(yt, yp) -> Any:
from keras import backend as K
return K.sum(K.cast((1 - yp) * (0 + yt) > 0.5, 'float')) / K.sum(0 + yt)
return K.sum(K.cast((1 - yp) * (0 + yt) > 0.5, 'float')) / K.maximum(1.0, K.sum(0 + yt))
def load_keras() -> Any: