AI Paper Atlas

Research map / Probabilistic and scientific modeling / Differential equations and PDEs

Decoupling Universal Laws and Environmental Heterogeneity: A Physics-Inspired Framework for Robust Spatio-Temporal Forecasting

Aoyu Liu, Liming Wei, Yaying Zhang

ICML 2026 · Probabilistic and scientific modeling › Differential equations and PDEs · approach: Physics-informed operator learning

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace the PDE solver with an MLP (ablation)STPDE · CA-D3 (ID and OOD)MAEHurt
add the Invariant Diffusion Operator (ablation)STPDE · CA-D3 and LA (ID and OOD)MAEImproved
add the Environment Basis Manifold (ablation)STPDE · CA-D3 and LA (ID and OOD)MAEImproved
add stochastic environment perturbation (ablation)STPDE · CA-D3 and LA (OOD)MAEImproved
use a GCN instead of the linear-attention Invariant Diffusion Operator (ablation)STPDE · CA-D3 and LAMAEHurt
increase hidden dimension D from 64 to 256, then 320 (ablation)STPDE · CA-D3, LAMAEMixed
raise perturbation ratio to about 25% or lower to 5% (ablation)STPDE · CA-D3, LAMAEHurt
increase number of environment bases K beyond 4 (ablation)STPDE · CA-D3, LAMAENo clear effect
use STPDE in place of strongest baseline for in-distribution forecastingID forecasting · LargeST four regionsMAE (averaged): 1.5% reductionImproved
use STPDE in place of strongest baseline for OOD generalizationOOD forecasting · LargeST four regionsMAE (averaged): 4.0% reductionImproved
adapt by reconstructing only the Environment Basis Manifold on 10% target dataSTPDE · CA-D3 few-shot transferMAE: CrossST 17.51 -> 16.25Improved
use OOD-specific methods (STONE, STOP) instead of ID-oriented GWNetOOD generalization baselines · LargeSTMAE: CA-D3 OOD: GWNet 16.84, STONE 17.89, STOP 17.68Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy