Model-selective causal estimators (IPW, regression) can jump discontinuously under small data shifts because they pick one SCM from a multimodal posterior, while effect-aligned summaries (posterior mean/median) remain continuous.
arXiv preprint arXiv:1904.02826 , year=
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
representative citing papers
A single numerical Jacobian, under a log-linear monomial assumption, recovers global identifiable parameter combinations and supports likelihood-based uncertainty quantification.
A conceptual argument that data-driven scientific models generalize outside their training data only when their mathematical form matches the true governing equation, which rules out neural operators and neural boundary-value models.
citing papers explorer
-
Two Layers of Instability in Causal Estimation
Model-selective causal estimators (IPW, regression) can jump discontinuously under small data shifts because they pick one SCM from a multimodal posterior, while effect-aligned summaries (posterior mean/median) remain continuous.
-
Invariant Image Reparameterisation: Bridging Symbolic and Numerical Methods for Identifiability Analysis, Model Reduction, and Prediction
A single numerical Jacobian, under a log-linear monomial assumption, recovers global identifiable parameter combinations and supports likelihood-based uncertainty quantification.
-
From inverse problems to neural operators: prediction, mechanism, and generalization of data-driven models
A conceptual argument that data-driven scientific models generalize outside their training data only when their mathematical form matches the true governing equation, which rules out neural operators and neural boundary-value models.