U-Mixer: An Unet-Mixer Architecture with Stationarity Correction for Time Series Forecasting

Time series forecasting is a crucial task in various domains. Caused by factors such as trends, seasonality, or irregular fluctuations, time series often exhibits non-stationary.