Guiding Spectral Grouping

Background: The effectiveness of a modular plugin designed to enhance multivariate time series forecasting models by modeling fine-grained, frequency-aware channel-patch dependencies.

Question / Future Work: Future work should explore incorporating external domain knowledge, such as meteorological or physiological priors, to guide the spectral grouping and filtering process used by the model.