Bayesian fairness

Typ
Examensarbete för masterexamen
Program
Publicerad
2020
Författare
Belfrage, Amanda
Berg Marklund, David
Modellbyggare
Tidskriftstitel
ISSN
Volymtitel
Utgivare
Sammanfattning
This thesis aims to extend the Bayesian fairness algorithm created by Dimitrakakis et al. to be able to handle continuous data. Using bagging to approximate the data we aim to reduce the problem to a computable task that still performs well enough to be an improvement over using the true underlying data. Even though promising results where found for using bagging with discrete data, the continuous version of the algorithm did not work.
Beskrivning
Ämne/nyckelord
Fairness, Bayesian fairness, algorithm
Citation
Arkitekt (konstruktör)
Geografisk plats
Byggnad (typ)
Byggår
Modelltyp
Skala
Teknik / material