Stationarity-Aware Retrieval-Augmented Time Series Forecasting

Arxiv CS.LG (machine learning)

The paper proposes a RAG-inspired approach for time series forecasting that handles non-stationarity and regime shifts by retrieving relevant historical patterns. The method aims to improve fully parametric forecasters by augmenting them with retrieved examples.

rag

Arxiv CS.LG (machine learning)·Shiqiao ZhouHolger Sch\"onerZipeng WuEdouard Fouch\'eIAG WilsonShuo Wang

7 days ago