Allow BatchBALD to not consider some samples #62
Merged
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
This is useful for active learning in that samples that are already labeled should not be selected for (re-)labeling. Note that the implementation computes confidence scores for these excluded samples, just as it does for all other samples that are not selected.
Leengit
changed the title
Leengit
force-pushed
the
batchbald_excluded_samples
branch
from
66fedba to
9c6e9d6
Compare
|
For source-code reviewing, it may be best to focus on the first commit, which shows the code changes in their original location. The second commit moves a bunch of code into a subroutine as the original routine was getting too long, which makes |
|
Thanks for keeping the commits separate to make it easier to review. |
manthey
approved these changes
Aug 26, 2024
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Outdated suggestions cannot be applied.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.
For reasons of efficiency,
BatchBALDdoes not give aBatchBALDscore for every sample. Instead it finds and scores the best samples. The remaining samples are given larger scores (higher certainty) based upon a fallback computation, which is currently a possibly shiftedconfidencescore.However, sometimes (e.g., when a sample has already been labeled by the user) we don't want BatchBALD to waste one of its "best-sample" slots on some subset of the samples. This is now implemented via a new
ComputeCertainty.set_batchbald_excluded_samplesmethod. The specified samples will never be selected as best samples and will instead always receive a possibly shiftedconfidencescore.This is a needed step in addressing DigitalSlideArchive/superpixel-classification#18.