发表机构
Tsinghua University; ByteDance（清华大学; 字节跳动）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文提出 COSTA 联合伯努利设计，通过阈值化高斯克罗内克参数化优化网络-时间干扰下的协方差，结合图-ψ中心极限等理论，在 RetailRocket 等数据集实验中显著降低 RMSE，实现可靠因果推断。
AI 中文摘要
针对随时间观测的网络实验面临网络溢出、时间 carryover 以及由设计刻意引入的依赖问题，本文提出 COSTA——协方差优化时空处理分配（Covariance-Optimized Spatiotemporal Treatment Allocation），一种针对单元-时间分配的联合伯努利设计。在常规处理边际和非负线性网络-时间暴露模型下，持续全处理组与全对照组对比的 Horvitz-Thompson（HT）偏差恰好为分配切割的负期望权重；协方差水平的方差包络可产生均方误差（MSE）界，该界可直接针对分配协方差进行优化。为扩展该设计，本文引入阈值化高斯克罗内克参数化，其镜像网络与时间暴露算子，同时保留有效的伯努利边际。接下来，本文针对由设计处理依赖与干扰诱导的结果依赖的联合效应发展推断理论：生成分离 HT 贡献的潜在块间的典型相关系数，提供图-ψ中心极限与网络-HAC理论所需系数；谱底与远行质量条件给出基础充分检验。该框架涵盖稀疏、块、克罗内克、局部分解及其他满足这些条件的结构化协方差序列。半合成 RetailRocket 与 MovieLens 实验显示，在默认设置下，对线性、非线性及需求替代结果曲面，均实现显著的均方根误差（RMSE）降低，且模型辅助的以设计为中心的区间校准良好。
英文摘要
Experiments on networks observed over time face network spillovers, temporal carryover, and dependence deliberately introduced by the design. We propose COSTA---Covariance-Optimized Spatiotemporal Treatment Allocation---a joint Bernoulli design for unit--time assignments. Under common treatment marginals and a nonnegative linear network--temporal exposure model, Horvitz--Thompson bias for the sustained all-treated versus all-control contrast is exactly the negative expected weight of an assignment cut. A covariance-level variance envelope yields an MSE bound that can be optimized directly over assignment covariance. To scale this design, we introduce a thresholded-Gaussian Kronecker parameterization that mirrors the network and temporal exposure operators while preserving valid Bernoulli marginals. We next develop inference theory for the joint effects of designed treatment dependence and interference-induced outcome dependence. Canonical correlations between latent blocks generating separated HT contributions supply the coefficients required by graph-$ψ$ central limit and network-HAC theory; a spectral-floor and far-row-mass condition gives a primitive sufficient check. The framework covers sparse, block, Kronecker, locally factored, and other structured covariance sequences satisfying these conditions. Semi-synthetic RetailRocket and MovieLens experiments show substantial default-setting RMSE reductions and well-calibrated model-assisted design-centered intervals across linear, nonlinear, and demand-substitution outcome surfaces.
