1 paper · 1 filter
Yixuan Zhang, Jiabin Luo, Zhenggang Wang +2
Fairness concerns are increasingly critical as machine learning models are deployed in high-stakes applications. While existing fairness-aware methods typically intervene at the mo…