[ML] Pseudo-Huber loss function #1168
Merged
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
tveasey
reviewed
Apr 30, 2020
tveasey
reviewed
May 4, 2020
tveasey
approved these changes
May 4, 2020
This reverts commit 003801d.
…-loss-function
tveasey
pushed a commit
to tveasey/ml-cpp-1
that referenced
this pull request
May 5, 2020
This PR implements Pseudo-Huber loss function and integrates it into the RegressionRunner. Since it has a parameter, I needed to reimplement the persist and restore functionality in order to be able to save the state of the loss functions (the same functionality is useful for MSLE and multiclass classification). I also did some refactoring of unit tests avoid code duplication.
Merged
valeriy42
added a commit
to valeriy42/ml-cpp
that referenced
this pull request
May 6, 2020
This PR implements Pseudo-Huber loss function and integrates it into the RegressionRunner. Since it has a parameter, I needed to reimplement the persist and restore functionality in order to be able to save the state of the loss functions (the same functionality is useful for MSLE and multiclass classification). I also did some refactoring of unit tests avoid code duplication.
This was referenced May 6, 2020
valeriy42
added a commit
that referenced
this pull request
May 6, 2020
* [ML] Pseudo-Huber loss function (#1168) This PR implements Pseudo-Huber loss function and integrates it into the RegressionRunner. Since it has a parameter, I needed to reimplement the persist and restore functionality in order to be able to save the state of the loss functions (the same functionality is useful for MSLE and multiclass classification). I also did some refactoring of unit tests avoid code duplication. * Fix test threshold (#1195) The unit test CBoostedTreeTest/testPiecewiseConstant was failing on Linux only. This PR adjusts the test threshold slightly. Co-authored-by: David Roberts <dave.roberts@elastic.co>
valeriy42
added a commit
that referenced
this pull request
May 6, 2020
* [ML] Pseudo-Huber loss function (#1168) This PR implements Pseudo-Huber loss function and integrates it into the RegressionRunner. Since it has a parameter, I needed to reimplement the persist and restore functionality in order to be able to save the state of the loss functions (the same functionality is useful for MSLE and multiclass classification). I also did some refactoring of unit tests avoid code duplication. * Fix test threshold (#1195) The unit test CBoostedTreeTest/testPiecewiseConstant was failing on Linux only. This PR adjusts the test threshold slightly. Co-authored-by: David Roberts <dave.roberts@elastic.co>
valeriy42
added a commit
that referenced
this pull request
May 8, 2020
While adding the additional function parameter in #1168, I wired it in the constructor of the MSLE loss function, but not in the computation of the objective. This PR fixes this, it basically substitutes log(1+x) by log(offset+x) in many different places. I mark it as a non-issue since the MSLE loss function was not released yet.
valeriy42
added a commit
to valeriy42/ml-cpp
that referenced
this pull request
May 8, 2020
While adding the additional function parameter in elastic#1168, I wired it in the constructor of the MSLE loss function, but not in the computation of the objective. This PR fixes this, it basically substitutes log(1+x) by log(offset+x) in many different places. I mark it as a non-issue since the MSLE loss function was not released yet.
valeriy42
added a commit
to valeriy42/ml-cpp
that referenced
this pull request
May 8, 2020
While adding the additional function parameter in elastic#1168, I wired it in the constructor of the MSLE loss function, but not in the computation of the objective. This PR fixes this, it basically substitutes log(1+x) by log(offset+x) in many different places. I mark it as a non-issue since the MSLE loss function was not released yet.
valeriy42
added a commit
that referenced
this pull request
May 11, 2020
While adding the additional function parameter in #1168, I wired it in the constructor of the MSLE loss function, but not in the computation of the objective. This PR fixes this, it basically substitutes log(1+x) by log(offset+x) in many different places. I mark it as a non-issue since the MSLE loss function was not released yet.
valeriy42
added a commit
that referenced
this pull request
May 11, 2020
While adding the additional function parameter in #1168, I wired it in the constructor of the MSLE loss function, but not in the computation of the objective. This PR fixes this, it basically substitutes log(1+x) by log(offset+x) in many different places. I mark it as a non-issue since the MSLE loss function was not released yet.
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 PR implements Pseudo-Huber loss function and integrates it into the RegressionRunner. Since it has a parameter, I needed to reimplement the persist and restore functionality in order to be able to save the state of the loss functions (the same functionality is useful for MSLE and multiclass classification). I also did some refactoring of unit tests avoid code duplication.
Relates to #973