发表机构
McGill University; University of Illinois Chicago; Mila – Quebec AI Institute; Norwegian University of Science and Technology; University of California, Santa Barbara; NVIDIA Corporation; University of British Columbia（麦吉尔大学; 伊利诺伊大学芝加哥分校; 米拉-魁北克人工智能研究所; 挪威科技大学; 加利福尼亚大学圣巴巴拉分校; 英伟达公司; 不列颠哥伦比亚大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
该研究提出保留来源的多视图融合方法 PACT，用于人机协作的类型化动作接纳，经多组实验验证其在风险覆盖等指标上的优势，明确一致性需来源支持才构成印证。
AI 中文摘要
对于具身系统而言，仅预测一致性不足以确定证据是否支持采取行动，证据的来源至关重要。对同一观测结果的重复推断会在不增加证据的情况下提升一致性，而源局部值无法揭示输出是否具有可单独计数的来源。PACT 将证据可计数性视为保留来源的融合与类型化动作接纳的关系变量。提供的来源划分定义了可计数单元，PACT 保留每个单元内共享的坐标支持，仅跨单元累积，并将未满足的释放条件映射为保持、确认或 fallback。在所述假设下，源局部值无法识别可计数性；坐标 meet 是满足单例保真和插入非放大的最大预算，具有粗化单调性和固定划分稳定性。在 48 个场景集群的 31200 次评估中，PACT 达到共同支持归一化风险覆盖面积（ncsAURC）为 0.0861。排除构建的共识臂后，来源划分聚合相比单例聚合使 ncsAURC 降低 0.0557，而印证对比消失。在完整来源记录上，原生评分偏向 PACT，但共同后验峰值评分缩小了其与嵌套 Dirichlet 和乘积融合的差异。在不变预测上重新分配来源会按预期移动证据预算。在离线人机协作中，八倍相机内重复使每个检查点的 720 个类型化响应保持不变；按相机分组的 PACT 在 60 个回合中接纳了 57 个 Qwen3-VL-32B 参考一致候选中的 47 个，未观察到参考不一致的接纳。PACT 将计算多重性与证据多重性分离：仅当来源允许单独累积时，一致性才构成印证。
英文摘要
Better probability scores do not establish that evidence has been counted correctly. Repeated inference over one observation can improve predictions without adding an evidential origin. Source-local numerical attributes alone cannot in general distinguish repeated derivations from separately countable acquisitions. PACT (Provenance-Aware evidence Conservation and Typed action admission) separates evidence magnitude from countability through a supplied provenance partition. Under singleton fidelity and insertion non-amplification, the coordinatewise meet is the unique pointwise greatest admissible within-component rule. Component budgets add under stated commensurability and separate-component additivity assumptions. Matched reassignments hold numerical outputs fixed while varying the counting relation. In four of 12 replicated-source tests on HandWritten, false refinement lowers macro-averaged negative log-likelihood and Brier score while increasing normalized common-support area under the risk-coverage curve (ncsAURC). In the controlled handover benchmark, removing the constructed adversarial-consensus condition leaves a 0.056 reduction in ncsAURC for provenance-partition aggregation relative to singleton aggregation under the same score functional. The corroboration contrast disappears, and method ranking remains selection-score dependent. In offline, reference-based human-robot collaboration with four prompts per camera and all other admission inputs fixed, duplicating each prompt output within its camera from multiplicity one to eight leaves all 720 PACT typed responses per checkpoint unchanged. Probability quality and evidence countability require separate evaluation.
Comments35 pages, 8 figures, 15 tables. Revised manuscript with clarified theoretical assumptions and evaluation scope. Code and supporting materials: https://github.com/ZekaiJ/PACT
