实用 DML 与 AI 学习表示
Pragmatic DML with AI-Learned Representations
Andres Aradillas Fernandez · Victor Chernozhukov · Carlos Cinelli · Sven Klaassen · Whitney Newey · Martin Spindler · et al.
中文摘要
文本、图像等丰富对象越来越多地通过 AI 学习表示作为控制变量进入因果分析。对一大类估计量而言,用表示替代原始对象会使目标偏移,偏移量为两个误差的内积:一个来自结果回归,一个来自平衡权重。因此偏差受二者规模乘积的约束。由此可得三点结论。表示固定时,常规交叉拟合去偏机器学习(DML)区间对表示特定目标有效;当近似偏差相对抽样误差可忽略时,该区间对因果目标也有效。若该偏差不可忽略,敏感性分析可界定因果目标并为端点提供置信界。当嵌入在样本上调优时,三种方法可保持有效推断:固定菜单微调、联合损失调优和双重表示选择。一个多模态需求应用展示了该工作流程。
关键要点
- 01用 AI 学习表示替代原始对象作控制变量,会使因果目标偏移为结果回归误差与平衡权重误差的内积。
- 02表示固定时,交叉拟合 DML 区间对表示特定目标有效;近似偏差可忽略时对因果目标也有效。
- 03偏差不可忽略时,敏感性分析可界定因果目标并为端点给出置信界。
- 04嵌入在样本上调优时,固定菜单微调、联合损失调优和双重表示选择可保持有效推断。
- 05多模态需求应用展示了该工作流程。
解读
尚无解读。
原始英文摘要
arXiv:2610.01935v3 Announce Type: replace-cross Abstract: Text, images, and other rich objects increasingly enter causal analysis through AI-learned representations used as controls. For a broad class of estimands, replacing the raw object by its representation shifts the target by the inner product of two errors: one in the outcome regression and one in the balancing weight. The bias is therefore bounded by the product of their sizes. Three consequences follow. With a fixed representation, the usual cross-fitted debiased machine learning (DML) interval is valid for a representation-specific target, and for the causal target when the approximation bias is negligible relative to sampling error. When this bias may not be negligible, a sensitivity analysis bounds the causal target and supplies confidence bounds for the endpoints. When the embedding is tuned on the sample, three approaches preserve valid inference: fixed-menu fine-tuning, joint-loss tuning, and double representation selection. A multimodal demand application illustrates the workflow.