面向智能体路由的可审计早停:Local DP 下的账本验证逐次运行证书
Auditable Early Stopping for Agentic Routing: Ledger-Verified Run-Wise Certificates under Local DP
AI总结:
研究工具使用智能体 best-first 路由在 Local DP 下的可审计早停问题,提出逐次运行证书,通过账本记录与指数竞赛耦合密钥实现安全停止,并在合成图与真实流水线上验证了低开销与确定性重放。
摘要:
我们研究工具使用智能体的 best-first 路由何时能在不遗漏更优叶子节点的情况下停止探索,同时保持 local differential privacy (LDP) 并留下审计轨迹。我们引入一种逐次运行证书,将每个节点的密钥耦合到实现叶子扰动的同一场指数竞赛中;常规的停止规则(当 $F$ 中所有 $v$ 的 Key$(v)$ 最大值 $\le B^*$ 时停止)随后对实际运行进行认证。我们在具有子节点划分的上下文索引前缀 DAG 上提供两种认证模式:(i) Exact(已知计数),采用带胜者复用的惰性偏移传播;以及 (ii) Surrogate(仅上界),将密钥锚定于父级代理竞赛,并允许通过 $κ= \log(N / N_{ub})$ 进行验证器收紧。一个小型编译器强制执行划分属性,且可容许的、独立于竞赛的 M(tau) 保持密钥可靠。账本记录均匀分布、计数和平局处理;隐私性通过后处理得以保障。在合成图和一个小型真实流水线上的实验表明,停止紧凑、重放确定且开销低。
英文摘要:
We address when a best-first router for tool-use agents can stop exploring without missing a better leaf, while preserving local differential privacy (LDP) and leaving an audit trail. We introduce a run-wise certificate that couples each node's key to the same exponential race that realizes leaf perturbations; the usual halting rule (stop when the maximum over $v$ in $F$ of Key$(v) \le B^*$) then certifies the realized run. We give two certified modes on context-indexed prefix DAGs with child partition: (i) Exact (known counts), using lazy offset propagation with winner reuse; and (ii) Surrogate (upper bounds only), which anchors keys to a parent-level surrogate race and allows validator tightening via $κ= \log(N / N_{ub}$). A small compiler enforces the partition property, and an admissible, race-independent M(tau) keeps keys sound. The ledger logs uniforms, counts, and tie handling; privacy follows by post-processing. Experiments on synthetic graphs and a small real pipeline show tight stopping, deterministic replay, and low overhead.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。