[Review] Print debug message if SVM convergence is poor #3562
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
tfeher
added
improvement
Codecov Report
@@ Coverage Diff @@
## branch-0.19 #3562 +/- ##
===============================================
- Coverage 80.73% 80.71% -0.03%
===============================================
Files 227 227
Lines 17605 17619 +14
===============================================
+ Hits 14214 14221 +7
- Misses 3391 3398 +7
Flags with carried forward coverage won't be shown. Click here to find out more.
Continue to review full report at Codecov.
|
|
rerun tests |
|
@gpucibot merge |
|
rerun tests |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
closes #947
If the input data for SVM is not normalized correctly, then convergence can be very slow. The solver can even fail to converge. This PR detects such cases and prints a debug message with suggestions how to fix this problem.
Such problems were reported in #947, #1664, #2857, #3233. The threshold for reporting is set so that the message is printed in those cases. I have tested several properly normalized cases to confirm that the message is not shown. Still, the threshold for printing the message does not have a proper theoretical justification, and false positives might occur. Therefore only a debug message is shown instead of a warning.