发表机构
Tsinghua University（清华大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对无历史数据的在线因果发现，提出FOCUS算法，通过序贯自适应干预和可计算集中不等式，在固定置信度下高效恢复因果DAG及边权重，并匹配理论下界。
AI 中文摘要
我们研究完全在线的固定置信度因果发现，且不依赖任何历史观测数据。从零样本开始，学习器在线性高斯结构方程模型下，序贯地选择干预以恢复因果DAG及其边权重。我们为任何$(\epsilon,\delta)$-正确的算法建立了实例相关的下界，并提出了FOCUS，它通过在线最大-最小博弈自适应地分配干预。一个关键贡献是针对累积KL散度的可计算集中不等式，该散度涉及跨干预共享的因果参数。我们证明了FOCUS是$(\epsilon,\delta)$-正确的，并且其期望停止时间在$\Theta(\log(1/\delta))$依赖项上匹配下界，直至一个实例相关的常数。实验展示了改进的结构和边权重恢复，并确认了预测的停止时间趋势。我们的代码可在该https URL上获取。
英文摘要
We study fully online fixed-confidence causal discovery without any historical observational data. Starting from zero samples, the learner sequentially selects interventions to recover both the causal DAG and its edge weights under a linear-Gaussian structural equation model. We establish an instance-dependent lower bound for any $(ε,δ)$-correct algorithm and propose \textsc{FOCUS}, which adaptively allocates interventions through an online max--min game. A key contribution is a computable concentration inequality for the accumulated KL divergence involving causal parameters shared across interventions. We prove that \textsc{FOCUS} is $(ε,δ)$-correct and that its expected stopping time matches the lower bound in its $Θ(\log(1/δ))$ dependence up to an instance-dependent constant. Experiments demonstrate improved structure and edge-weight recovery and confirm the predicted stopping-time trend. Our codes are available on https://anonymous.4open.science/r/FOCUS_code-76E5
