Add adaptive stdev scaling to local search neighborhood sampling #954#1303
Open
JannisKst wants to merge 2 commits intodevelopmentfrom
Open
Add adaptive stdev scaling to local search neighborhood sampling #954#1303JannisKst wants to merge 2 commits intodevelopmentfrom
JannisKst wants to merge 2 commits intodevelopmentfrom
Conversation
mwever
reviewed
Mar 10, 2026
|
|
||
| hp_names = list(start_points[0].config_space.keys()) | ||
|
|
||
| # Initial standard deviation used when sampling continuous neighbors. |
There was a problem hiding this comment.
Please, no magic constants in our code :-)
Can you please add these parameters as actual parameters for this class? You can still set default values, no problem, but these parameters should not just simply pop up within an internal method.
| improvement_count[i] = 0 | ||
|
|
||
| # Increase exploration radius if the search improves consistently. | ||
| if improvement_count[i] >= 3 and n_no_plateau_walk[i] == 0: |
There was a problem hiding this comment.
Also here please no magic constants in between. The limit for scaling out or zooming in should be a parameter of this class.
| n_no_plateau_walk[i] += 1 | ||
| # Reduce exploration radius during plateau walking to refine the | ||
| # search locally around the current candidate. | ||
| if n_no_plateau_walk[i] % 3 == 0: |
`_stdev_init`, `_stdev_min`, `_stdev_max`, `_upscale_thresh`, `_downscale_interval` - `_stdev_max` can be None, in which case it is computed heuristically
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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 adds adaptive scaling of the neighborhood sampling standard deviation for continuous hyperparameters based on local search progress.
The scaling follows these rules:
These changes were benchmarked in the SMAC3-Benchmarking repository. The repository also contains additional details explaining the design and implementation choices.