Table 6.

Modelling stage and different technical solutions for algorithm bias and discrimination.

stage	technical solution
pre-processing	reweighing subjects [98] oversampling minority groups [99] disparate impact remover [72] learning fair representations [100]
in-processing	adversarial debiasing [101] fairness constraint [73,102] counterfactual fairness [103]
post-processing	calibrated equality of odds [104] reject option classification [98]