[WIP] Potential enhancement to multiplicative updating #33
Conversation
This file contains hidden or 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
ejolly
changed the title
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Outdated suggestions cannot be applied.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.
This tries to use the approach in Blondel et al and Zhu, 16 for using a weight/mask matrix to ignore missing values while doing the Lee & Seung multiplicative update. The latter implementation can be found in a git repo here
It initially seemed to work well for some data where out-of-sample performance increased, but for other data it performed horrendously where predictions weren't even on the same scale as the user ratings. I got a little excited and merged it into master in #31, but given how finicky it seems to be, I've unmerged it now and I'm leaving it in this branch/PR for posterity.