<ul><li>Multivariable time series forecasting methods with exogenous variables enhance prediction accuracy.</li><li>A novel architecture called Sonnet combines learnable wavelet transformations and spectral analysis for forecasting.</li><li>Sonnet outperforms competitive baselines on 34 out of 47 forecasting tasks, with an average MAE reduction of 1.1%.</li><li>Integrating Multivariable Coherence Attention improves forecasting models, reducing MAE by 10.7% on average in challenging tasks.</li></ul>

Sonnet: Spectral Operator Neural Network for Multivariable Time Series Forecasting

Discover more