Learning from weak, proxy, or relative supervision is common when ground-truth labels are unavailable, but robustness under distribution shift remains poorly understood because the supervision mechanism itself may change across environments. We formalize this phenomenon as supervision drift, defined as changes in $P(y \mid x, c)$ across contexts, and study it in CRISPR-Cas13d transcriptomic perturbation experiments where guide efficacy is inferred indirectly from RNA-seq responses. Using publicly available data spanning two human cell lines and multiple post-induction timepoints, we construct a controlled non-IID benchmark with explicit domain (cell line) and temporal shifts, while reusing a fixed weak-label construction across all contexts to avoid changing targets. Across linear and tree-based models, weak supervision supports meaningful learning in-domain (ridge $R^2 = 0.356$, Spearman $ρ= 0.442$) and partial cross-cell-line transfer ($ρ\approx 0.40$). In contrast, temporal transfer collapses across all model classes considered, yielding negative $R^2$ and weak or near-zero $ρ$ (ridge $R^2 = -0.145$, $ρ= 0.008$; XGBoost $R^2 = -0.155$, $ρ= 0.056$; random forest $R^2 = -0.322$, $ρ= 0.139$). Additional robustness analyses using externally recomputed weak labels, shift-score quantification, and simple mitigation baselines preserve the same qualitative pattern. Feature-label association and feature-importance analyses remain relatively stable across cell lines but change sharply over time, indicating that failures arise from supervision drift rather than model capacity or simple covariate shift. These results show that strong in-domain performance under weak supervision can be misleading and motivate feature stability as a lightweight diagnostic for non-transferability before deployment.

翻译：当真实标签不可用时，通过弱监督、代理监督或相对监督进行学习很常见，但在分布偏移下的鲁棒性仍未被充分理解，因为监督机制本身可能在不同环境中发生变化。我们将此现象形式化为监督漂移，即上下文间$P(y \mid x, c)$的变化，并在CRISPR-Cas13d转录组扰动实验中加以研究，其中引导效率通过RNA-seq响应间接推断。利用跨越两种人类细胞系和多个诱导后时间点的公开可用数据，我们构建了一个受控制的非独立同分布基准，包含显式域（细胞系）和时间偏移，同时在所有上下文中复用固定的弱标签构建以避免目标变化。在线性和基于树的模型中，弱监督支持有意义的域内学习（岭回归$R^2 = 0.356$，斯皮尔曼$ρ= 0.442$）和部分跨细胞系迁移（$ρ\approx 0.40$）。相反，所有考虑模型类的时间迁移均失效，产生负$R^2$和弱或接近零的$ρ$（岭回归$R^2 = -0.145$，$ρ= 0.008$；XGBoost $R^2 = -0.155$，$ρ= 0.056$；随机森林$R^2 = -0.322$，$ρ= 0.139$）。使用外部重计算的弱标签、偏移分数量化及简单缓解基线的附加鲁棒性分析保持了相同的定性模式。特征-标签关联和特征重要性分析在细胞系间保持相对稳定，但随时间急剧变化，表明失败源于监督漂移而非模型容量或简单协变量偏移。这些结果表明，弱监督下的强域内性能可能具有误导性，并激发特征稳定性作为部署前非可迁移性的轻量级诊断指标。