STEPS reformulates test-time adaptation for time series forecasting as a Dirichlet boundary value problem on a temporal manifold and solves for smooth error corrections, yielding 26.82% average relative MSE reduction over zero-shot baselines.
arXiv preprint arXiv:2506.23424 , year=
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.LG 2years
2026 2verdicts
UNVERDICTED 2roles
background 1polarities
background 1representative citing papers
ADOWIP uses a decision-loss priority gate to update only when loss exceeds an empirical quantile under budget constraints, showing lower held-out decision loss than always-update or fixed-period baselines on ETT tasks.
citing papers explorer
-
STEPS: A Temporal Smooth Error Propagation Solver on the Manifolds for Test-Time Adaptation in Time Series Forecasting
STEPS reformulates test-time adaptation for time series forecasting as a Dirichlet boundary value problem on a temporal manifold and solves for smooth error corrections, yielding 26.82% average relative MSE reduction over zero-shot baselines.
-
Adapt Only When It Pays: Budgeted Decision-Loss Priority for Delayed Online Time-Series Adaptation
ADOWIP uses a decision-loss priority gate to update only when loss exceeds an empirical quantile under budget constraints, showing lower held-out decision loss than always-update or fixed-period baselines on ETT tasks.