Accurate Parameter-Efficient Test-Time Adaptation for Time Series Forecasting
This addresses the issue of high memory and compute costs in test-time adaptation for time series forecasting, offering an incremental improvement for practitioners needing efficient model updates.
The paper tackles the problem of non-stationary time series degrading pre-trained forecasting models by proposing PETSA, a parameter-efficient test-time adaptation method that updates only small calibration modules, achieving competitive or better performance across all horizons on benchmark datasets.
Real-world time series often exhibit a non-stationary nature, degrading the performance of pre-trained forecasting models. Test-Time Adaptation (TTA) addresses this by adjusting models during inference, but existing methods typically update the full model, increasing memory and compute costs. We propose PETSA, a parameter-efficient method that adapts forecasters at test time by only updating small calibration modules on the input and output. PETSA uses low-rank adapters and dynamic gating to adjust representations without retraining. To maintain accuracy despite limited adaptation capacity, we introduce a specialized loss combining three components: (1) a robust term, (2) a frequency-domain term to preserve periodicity, and (3) a patch-wise structural term for structural alignment. PETSA improves the adaptability of various forecasting backbones while requiring fewer parameters than baselines. Experimental results on benchmark datasets show that PETSA achieves competitive or better performance across all horizons. Our code is available at: https://github.com/BorealisAI/PETSA