Add FARE recourse method. #199
Open
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
* Add suitable pytest with a MockEnv * Add fare_actions_factory to easily generate action sets * Add sml-unitn/recourse-fare as dependency [1] De Toni, Giovanni, Bruno Lepri, and Andrea Passerini. "Synthesizing explainable counterfactual policies for algorithmic recourse with program synthesis." Machine Learning (2023): 1-21.
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Outdated suggestions cannot be applied.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 is an initial attempt to add the following two novel recourse methods to the library, FARE and E-FARE [1]. Differently from other approaches, FARE couples a reinforcement learning agent and a discrete search procedure, MCTS, to efficiently discover counterfactual interventions.
Currently, I have added FARE and a simple pytest to show its usage. Some class/function documentation is still missing, but I would like to start a discussion about what would be the best strategy to proceed.
This PR also adds as a dependency the sml-unitn/recourse-fare repository where the original code is located (in order to avoid bloating CARLA with additional files). Obviously, this can be changed.
Minor additions:
[1] De Toni, Giovanni, Bruno Lepri, and Andrea Passerini. "Synthesizing explainable counterfactual policies for algorithmic recourse with program synthesis." Machine Learning (2023): 1-21. doi:10.1007/s10994-022-06293-7