Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add counterfactual generator support for EvoTrees #225

Closed
5 tasks
RaunoArike opened this issue Jun 9, 2023 · 0 comments
Closed
5 tasks

Add counterfactual generator support for EvoTrees #225

RaunoArike opened this issue Jun 9, 2023 · 0 comments
Labels
could have This feature could be implemented during the software project difficult This is expected to be difficult. enhancement New feature or request

Comments

@RaunoArike
Copy link
Contributor

Though resolving #206 included implementing an interface to the EvoTreeClassifier from the MLJ model registry, support for generating counterfactuals for this model has not been implemented yet. The steps to take are the following:

  • As noted by the client in Interface to <: Supervised MLJ models #69, it might not be possible to retrieve the gradient information of MLJ models using Zygote.jl. Whether that is indeed the case should be investigated.
  • If Zygote.jl cannot be used for gradient access, alternatives should be explored. One possible alternative is ForwardDiff.jl, but this may lead to other problems, such as gradients being ill-defined due to the counterfactual loss function not being smooth.
  • If a working solution is found, implement it.
  • Add support for using the Feature Tweak generator on EvoTrees. This should be doable even if a working solution is not found for retrieving gradient information, as Feature Tweak does not require gradient information.
  • Document the whole process and thoroughly test the implemented functionality.
@RaunoArike RaunoArike added enhancement New feature or request difficult This is expected to be difficult. could have This feature could be implemented during the software project students labels Jun 9, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
could have This feature could be implemented during the software project difficult This is expected to be difficult. enhancement New feature or request
Projects
None yet
Development

No branches or pull requests

2 participants