发表机构
Seoul National University; Asteromorph（首尔国立大学; Asteromorph）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对有限粒子扩散引导中VCG更新可能增加残差风险的问题，提出风险校准提议传输（RCPT），利用留一法残差校准更新比例，在多项任务中改善终端指标。
AI 中文摘要
推理时引导通过改变将噪声传输到数据的动力学，无需重新训练即可结合预训练的扩散专家或奖励。Feynman-Kac 校正通过重要性加权的序列蒙特卡洛（SMC）补偿提议不匹配，其有限粒子行为取决于提议。方差控制引导（VCG）通过拟合线性漂移校正以最小化经验对数权重率方差来改进该提议。尽管其总体最优值不会恶化残差方差，但有限粒子 VCG 几乎可以消除其拟合残差，同时将新状态上的残差风险增加数个数量级。由此产生的更新可能降低未加权生成的质量或加速粒子坍缩。我们证明中心化的 Feynman-Kac 速率是归一化的传输残差，并且期望的拟合外收益恰好是总体余量减去系数估计惩罚。在正则性假设下，Wasserstein 分析使用该残差限制了未加权提议的终端误差。这些结果推动了风险校准的提议传输（RCPT），它使用删除留一法残差来校准 VCG 更新的保留比例，不增加模型调用，仅增加少量线性代数开销。在二维棋盘分布、支架装饰、分子性质优化和类条件 CIFAR-10 生成上的实验证明，RCPT 能够从有害的拟合更新中恢复。在分子和图像领域，RCPT 减轻了有害的拟合更新，并相对于未校准的 VCG 改善了一系列终端指标。
英文摘要
Inference-time steering combines pretrained diffusion experts or rewards without retraining by changing the dynamics that transport noise to data. Feynman-Kac correction compensates for proposal mismatch through importance-weighted sequential Monte Carlo (SMC), whose finite-particle behavior depends on the proposal. Variance-controlling guidance (VCG) improves that proposal by fitting a linear drift correction to minimize empirical log-weight-rate variance. Although its population optimum cannot worsen residual variance, finite-particle VCG can nearly eliminate its fitting residual while increasing residual risk on new states by orders of magnitude. The resulting update can degrade unweighted generation or accelerate particle collapse. We show that the centered Feynman-Kac rate is the normalized transport residual and that expected out-of-fit benefit is exactly population headroom minus coefficient-estimation penalty. Under regularity assumptions, a Wasserstein analysis bounds the unweighted proposal's terminal error using this residual. These results motivate Risk-Calibrated Proposal Transport (RCPT), which uses deletion leave-one-out residuals to calibrate the retained fraction of the VCG update, adding no model calls and only small linear-algebra overhead. Experiments on 2D checker distributions, scaffold decoration, molecular property optimization, and class-conditional CIFAR-10 generation demonstrate recovery from harmful fitted updates. Across molecular and image domains, RCPT mitigates harmful fitted updates and improves a broad range of terminal metrics relative to uncalibrated VCG.
CommentsEarlier version accepted at NeurIPS 2026 Workshop on AI for Stochastic Dynamics (STODY)
