发表机构
College of New Media and Communication, Tianjin University; School of Artificial Intelligence, Tianjin University（天津大学新媒体与传播学院; 天津大学人工智能学院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
研究时间图生成模型中分布漂移问题，通过掩码流匹配损失分析其不可修复性，给出实验结果，证明基于观测的校正器效果不佳，预言机消除误差效果最好，外推比不处理还差。
AI 中文摘要
时间图的生成模型在演化网络的一段上进行训练，然后部署到下一段，在这个间隔中性能会严重下降。我们表明这种下降是可推导的、普遍的，且无法从观测中修复。掩码流匹配损失在无独立性假设下精确分解为不可约熵加上一个散度，其沿训练路径的导数对于训练期间罕见而部署时常见的结构为正，随着训练概率趋于零而发散。实验上这种权衡是指数为 -0.605 的幂律（\(R^2 = 0.9977\)），漂移提高了采样器的误差下限且不改变达到该下限所需的步数。我们证明任何基于过去观测可测量的校正器至少会留下它所跟踪统计量的条件方差，并且只有当\(\mu^2 > v(1 - 2\rho)\)时趋势外推才优于仅信任最后一次观测。一个预言机可消除60%的误差，基于观测的最佳校正器能恢复其中的5.7%，外推比不做任何处理更差。
英文摘要
Generative models of temporal graphs are trained on one stretch of an evolving network and deployed on the next, and they degrade badly in the gap. We show this degradation is derivable, general, and not fixable from observations. The masked flow-matching loss decomposes exactly, with no independence assumption, into an irreducible entropy plus a divergence whose derivative along the training path is positive precisely for structures rare during training and common at deployment, diverging as their training probability goes to zero. Empirically the trade-off is a power law with exponent $-0.605$ ($R^2=0.9977$), and drift raises the sampler's error floor without changing how many steps reach it: across seven well-powered conditions the drift-period marginal error varies by at most $6\%$ over a $50\times$ range of sampling budgets, while the floor sits $2.2\times$ to $34.3\times$ above the in-period floor. Because the deployment period is observed, correction looks like a matter of measurement. It is not. We prove that any corrector measurable with respect to past observations leaves at least the conditional variance of the statistic it tracks, and that trend extrapolation beats trusting the last observation only when $μ^2>v(1-2ρ)$. Both premises are measurable and both go the wrong way: the drift is trendless and mean-reverting, with a one-step innovation as large as the drift itself. An oracle removes $60\%$ of the error, the best observation-based corrector recovers $5.7\%$ of that, and extrapolation is strictly worse than doing nothing clever.
