TY  - RPRT
TI  - Variance-Optimal Control Variates for Learning with Black-box Feedback
AU  - Zihao Zhao
AU  - Shuhan Zhang
AU  - Kai Wang
PY  - 2026
UR  - https://arxiv.org/abs/2610.04766
ID  - 2610.04766
ER  -