[SPARK-2552][MLLIB] stabilize logistic function in pyspark #1493
Conversation
|
Jenkins, test this please. |
|
QA tests have started for PR 1493. This patch merges cleanly. |
|
Jenkins, retest this please. |
|
QA tests have started for PR 1493. This patch merges cleanly. |
|
Jenkins, retest this please. |
|
QA tests have started for PR 1493. This patch merges cleanly. |
|
QA results for PR 1493: |
There was a problem hiding this comment.
Better would be
prob = exp(margin)/(1 + exp(margin))because for very small probabilities this computation is essentially doing 1 - 1. For example:
>>> from math import exp
>>> margin = -40
>>> 1 - 1 / (1 + exp(margin))
0.0
>>> exp(margin)/(1 + exp(margin))
4.248354255291589e-18
>>>There was a problem hiding this comment.
Yes, that is definitely better. Could you submit a PR? We don't need a JIRA for small changes. Btw, please cache exp(margin) instead of computing it twice.
There was a problem hiding this comment.
Sure, pull request here! #1652
Thanks a lot! :-)
to avoid overflow in `exp(x)` if `x` is large. Author: Xiangrui Meng <meng@databricks.com> Closes apache#1493 from mengxr/py-logistic and squashes the following commits: 259e863 [Xiangrui Meng] stabilize logistic function in pyspark
4 participants
to avoid overflow in
exp(x)ifxis large.