AI & ML
Pragmatic DML with AI-Learned Representations
arXiv stat.ML0 views
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.
Source:https://arxiv.org/abs/2610.01935PreviousArbitrary-Accuracy Neural Approximation with Optimal Neuron Count and Optimal Bit ComplexityNextMixing Regularization for Average-Reward $Q$-Learning and $Q$-Planning
Related posts
The "Unified Format" Nightmare in Enterprise IT (Air-Gapped Rooms & Double Timesheets)
0dev.to Watercooler
Phân tích thống kê Inter Miami vs Columbus Crew: Bài toán hàng thủ trước chuỗi phong độ thăng hoa
0dev.to Data
Biến Động Hàng Thủ: Barcelona Quan Tâm Lisandro Martinez, Man Utd Nhắm Nhân Sự Từ Fulham
0dev.to Data
Real-world documents are dirtier than your test fixtures — what agents shipping data pipelines should learn
0dev.to Data
Thanh Hóa hạ gục Hoàng Anh Gia Lai 2-1: Giá trị từ pha lập công then chốt của Matheus Martins
0dev.to Data
EPMOS Daily World-Scan — Friday, November 20, 2026
0dev.to News