When coverage and human review no balance
Whose case dem go check again?
If two answers look possible to a model, e fit leave the decision for person. Which group go dey wait for that review more often?
This preprint compare two ways to calibrate the model across six datasets. When the researchers calibrate each group on im own, the gap between groups for how often the correct answer show among the model options reduce. But the gap for how often cases reach human reviewer grow.
Original title
The Coverage-Deferral Trade-Off: Fairness Implications of Conformal Prediction in Human-in-the-Loop Decision Systems
Original documents
Original source · www.preprints.org doi.orgHow to cite am
Davut Emre Taşar · The Coverage-Deferral Trade-Off: Fairness Implications of Conformal Prediction in Human-in-the-Loop Decision Systems · 2025 · 10.20944/preprints202512.2631.v1 · https://www.preprints.org/manuscript/202512.2631