Dual-branch module improves time series forecasting under changing conditions
DBCM: Dual-Branch Calibration Module for Test-Time Adaptation in Time Series Forecasting
Machine Learning
Summary
Time series forecasting often struggles when patterns change over time, causing errors in predictions. The authors noticed that these errors are sometimes related across different data channels, not just isolated to one. They created a new method called the Dual-Branch Calibration Module (DBCM) that adjusts forecasts by sharing information across channels while still allowing for some channel-specific adjustments. Tests on various datasets show this approach works better than earlier methods.
What this means in practice
- •For data science teams: Improve accuracy of time series forecasts in dynamic environments by updating models using recent data while adapting shared residual patterns.
- •For industrial process controllers: Enhance prediction reliability in complex multi-sensor systems by applying dual-branch calibration to adjust for correlated errors across sensors at runtime.
Authors
Jaebyeong Jeon, Kunwoo Kang, Sang Yup Lee, Kibok Lee
Abstract
Non-stationarity induces test-time distribution shifts in time series forecasting (TSF), resulting in forecast residuals between predictions and the corresponding ground truth. Recent works on test-time adaptation for TSF address this by introducing lightweight calibration modules around frozen forecasters, updated using observations that become incrementally available at test time. Existing approaches typically employ channel-wise adapters, implicitly assuming that forecast residuals are inherently channel-specific. However, we observe that forecast residuals often exhibit substantial shared structure across channels, suggesting that fully independent channel-wise parameterization may be redundant. Motivated by this observation, we propose the Dual-Branch Calibration Module (DBCM), which shares adaptation capacity across channels through a shared adapter while retaining limited channel-specific flexibility through a low-rank channel-wise adapter, further enhanced by channel mixing and nonlinear gated modulation. Extensive experiments show that DBCM consistently outperforms prior methods across diverse datasets, horizons, and backbones.