AI 中文总结
本文提出信任比量化匹配可靠性，并设计置信校准匹配策略，实验表明在低信任时弃权优于始终匹配。
AI 中文摘要
只匹配你能识别的几何；否则分散惩罚。我们通过信任比 tau = epsilon / gamma（估计不确定性除以谱间隔）来量化该决策。在线性-二次匹配响应下，对于所选前 top-r 部署子空间中的探针，估计投影仪与预言机投影仪之间的预言机相对漂移随 tau^2 缩放——在 Davis-Kahan 分离区域 tau < 1/2 内为 O(tau^2)，其实际效用取决于常数。置信校准匹配（CCM）将 tau 转化为策略——当 tau 小时为方向性，否则逐渐变为各向同性——阈值来自校准而非定理（匹配位于分离区域；软匹配主要是启发式）。实验展示了两种机制，包括 UCI HAR 嵌入，其中始终匹配在每个单元上都比弃权（不执行）更差。
英文摘要
Match only geometry you can identify; otherwise spread the penalty. We quantify that decision by the trust ratio tau = epsilon / gamma (estimation uncertainty over spectral separation). Under the linear-quadratic Matching response, oracle-relative drift between estimated and oracle projector matching scales as tau^2 for probes in the chosen top-r deployment subspace -- O(tau^2) in the Davis-Kahan separation region tau < 1/2, with practical usefulness depending on constants. Confidence-Calibrated Matching (CCM) turns tau into a policy -- directional when tau is small, progressively isotropic when not -- with thresholds from calibration, not from the theorem (match sits in the separation region; soft is mostly heuristic). Experiments show both regimes, including UCI HAR embeddings where always-match is worse than abstain on every cell.
Comments14 pages. Companion to arXiv:2604.21395 and arXiv:2605.22800
