Accurate forecasting of time series data is essential in many fields. However, real-world time series are often characterized by noise, non-stationarity and multiscale temporal dependencies, which ...