Prejudice remover is an in-processing technique that adds a discrimination-aware regularization term to the learning objective
Arguments
- eta
fairness penalty parameter
- sensitive_attr
name of protected attribute
- class_attr
label name
Examples
if (FALSE) {
# An example using the Adult Dataset
load_aif360_lib()
ad <- adult_dataset()
model <- prejudice_remover(class_attr = "income-per-year", sensitive_attr = "race")
model$fit(ad)
ad_pred <- model$predict(ad)
}