Marouane El Hadari

Learning to Difference: Adaptive Reversible Differencing (AdaRDiff) for Time Series Forecasting

AdaRDiff is proposed, a generalized differencing approach that uses learnable weights to simplify the series through weighted differencing with previous time instants, and attains state-of-the-art forecast accuracy across eight benchmarks spanning electricity, weather, traffic, and energy, at negligible parameter cost.