Learning Deployable Causal Action Geometry under Temporal Non-Stationarity

Key points

• Even when the one-step response is identifiable, unrestricted cross-time generalization remains fundamentally ill-posed.1
• We propose deployable causal action geometry: the action-response shape that remains after anchoring away action-independent level shifts on a selected link scale and profiling out positive amplitude variation.1
• Building on this target, we develop a two-stage orthogonal estimator that first obtains cross-fitted pilot estimates of anchored responses and then performs honest profiled selection over link, representation, and shape classes.1
• Experiments on synthetic data and Dominick's support the proposed target and its transfer behavior.1

Sources (1)

• [1]Learning Deployable Causal Action Geometry under Temporal Non-Stationarity
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 12, 12:06 PM

  Learning a response law for continuous decisions from observational time series is central to many real-world decision systems, yet temporal non-stationarity poses a fundamental challenge: as outcomes drift over time, naively pooled models can conflate action effects with shifting baselines and even reverse action rankings.

  Even when the one-step response is identifiable, unrestricted cross-time generalization remains fundamentally ill-posed.

Extractive summary: sentences quoted from the sources.