1 paper · 1 filter
Chengchang Liu, Zongqi Wan, Haishan Ye +1
We study stochastic zeroth-order optimization with decision-dependent distributions, where the sampling law depends on the current decision and only noisy function values are avail…