MemCast: Memory-Driven Time Series Forecasting with Experience-Conditioned Reasoning
This work addresses the lack of explicit experience accumulation in LLM-based time series forecasting, which is important for applications like decision-making, though it appears incremental as it builds on existing methods.
The authors tackled the problem of time series forecasting by proposing MemCast, a framework that learns from training data to create a hierarchical memory for experience-conditioned reasoning, resulting in consistent outperformance over previous methods in experiments on multiple datasets.
Time series forecasting (TSF) plays a critical role in decision-making for many real-world applications. Recently, LLM-based forecasters have made promising advancements. Despite their effectiveness, existing methods often lack explicit experience accumulation and continual evolution. In this work, we propose MemCast, a learning-to-memory framework that reformulates TSF as an experience-conditioned reasoning task. Specifically, we learn experience from the training set and organize it into a hierarchical memory. This is achieved by summarizing prediction results into historical patterns, distilling inference trajectories into reasoning wisdom, and inducing extracted temporal features into general laws. Furthermore, during inference, we leverage historical patterns to guide the reasoning process and utilize reasoning wisdom to select better trajectories, while general laws serve as criteria for reflective iteration. Additionally, to enable continual evolution, we design a dynamic confidence adaptation strategy that updates the confidence of individual entries without leaking the test set distribution. Extensive experiments on multiple datasets demonstrate that MemCast consistently outperforms previous methods, validating the effectiveness of our approach. Our code is available at https://github.com/Xiaoyu-Tao/MemCast-TS.