← Back to all news
arXiv Machine Learning By Md Khalid Hasan Sakib, Dristi Datta, Manoranjan Paul, Davina White

GeoDose-CP: Graph-Local Conformal Inference for Continuous-Treatment Earth Observation

Read the original on arXiv Machine Learning →

GeoDose-CP introduces a graph‑local conformal inference framework for estimating localized stochastic potential outcomes when dealing with continuous or mixed continuous‑atomic treatments in Earth observation data. The method jointly models intervention‑induced treatment shifts, outcome‑scale Jacobians, and spatial residual dependence, and includes exact weighted candidate inversion, a scalable sparse approximation, and a refusal mechanism for inadequate support. Evaluation on controlled experiments, MineDoseBench, and a multi‑mine study in New South Wales demonstrates high selective coverage and identifies limitations when longitudinal treatment data are unavailable.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Machine Learning.

Related stories

arXiv Machine Learning
Sep 25

Diverse Geometries, Frozen Weights: Robust Heterogeneous Treatment-Effect Estimation via Causal Expert Ensembles

The paper introduces GeoACE, a five‑expert framework for estimating heterogeneous treatment effects that blends a common anchor‑correction estimator with overlap‑aware and outcome‑guided geometries. The ensemble’s task‑level weights are learned from internal validation predictions, frozen before test evaluation, and applied to experts refitted on the full development data. Adding the outcome‑free, overlap‑aware expert O‑Phi‑ACE consistently improves performance across seven benchmarks, achieving the lowest average rank among 11 comparators.

By Ali Haghpanah Jahromi, Mohammad Taheri
benchmarkssafety
More like this →