2026
Parameter Decorrelation via Transition-Variance Alignment for Multivariate Time-series Forecasting
ICML 2026poster
Multivariate time-series forecasting (MTSF) learns from high-dimensional covariates with strong temporal dependence, periodic structure, and cross-variable correlations. While modern pipelines often mitigate non-stationarity through instance-wise normalization and decomposition, these interventions …