弱监督学习在分布偏移下可能失效，需警惕时间变化带来的标签漂移。
Learning Stable Predictors from Weak Supervision under Distribution Shift
- 通过控制实验分离细胞系与时间维度的分布偏移，构建非独立同分布基准
- 弱监督在同细胞系内表现良好（皮尔逊相关ρ=0.442），跨时间迁移全失败
- 特征重要性随时间剧变，提示监督信号漂移是模型失效主因
当真实标签不可获取时，常采用弱监督、代理或相对监督。然而，分布偏移下的鲁棒性仍不清晰，因为监督机制本身可能随环境变化。本文将此现象定义为监督漂移（supervision drift），即P(y∣x,c)在不同上下文中的变化。研究基于CRISPR-Cas13d转录组扰动实验，从RNA-seq响应间接推断引导剂效能。利用覆盖两种人源细胞系及多个诱导后时间点的公开数据，构建了含显式领域（细胞系）与时间偏移的受控非独立同分布基准，并在所有情境中保持固定的弱标签构造以避免目标变动。在线性与树模型上，弱监督在域内表现良好（岭回归R²=0.356，斯皮尔曼ρ=0.442），跨细胞系转移部分有效（ρ≈0.40）。相反，时间迁移在所有模型中均崩溃，出现负R²和接近零的ρ（岭回归R²=-0.145，ρ=0.008；XGBoost R²=-0.155，ρ=0.056；随机森林R²=-0.322，ρ=0.139）。使用外部重算弱标签、偏移得分量化及简单缓解基线的额外鲁棒性分析，仍维持相同定性模式。特征-标签关联与特征重要性在细胞系间相对稳定，但随时间剧烈变化，表明失败源于监督漂移而非模型能力或简单协变量偏移。结果表明，弱监督下强域内性能可能具有误导性，提示应以特征稳定性作为部署前非可迁移性的轻量级诊断工具。
原文摘要 · Abstract (English)
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.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。