Retrieval Is Not Enough: Refreshing Memory for Frozen Time-Series Forecasters

2026-10-06

AI Summary

This study addresses the accuracy limitations of frozen predictors caused by outdated retrieved memories and uncalibrated correction weights. To overcome these challenges, we propose FreshCast, a plug-and-play framework centered on memory freshness and residual alignment. By continuously refreshing non-parametric memories, generating predictions via relational kernel regression, and performing closed-form weight calibration, our method achieves online adaptive enhancement without retraining. Extensive experiments across seven benchmarks and ten architectures demonstrate that FreshCast reduces the mean squared error by 14.6% and 5.6% for short and long input contexts, respectively, significantly outperforming existing baselines.

Abstract

Retrieval-augmented time-series forecasting uses the continuations of historical segments similar to the current context as references for a forecaster. Most existing methods build the retrieval memory once from the training segment, leaving observations revealed after deployment unavailable as references, and generally do not calibrate how much the retrieved information should influence a frozen forecaster. We identify two key determinants of retrieval utility for a frozen forecaster: whether the history still reflects the current state, and whether the correction it induces aligns with the forecaster's residual errors, an alignment that can shift between validation and deployment when the memory becomes stale. We propose FreshCast, a plug-in retrieval framework that keeps the forecaster frozen, continuously updates a non-parametric memory with new observations, forms a memory forecast through relational kernel regression, and calibrates its weight in closed form on the validation segment. Under a simplified generative model, we characterize the optimal combination gain through the second-order relation between forecaster error and memory correction, and show that a sufficiently long look-back can make periodic memory information redundant. Across seven benchmarks and ten forecasting architectures, FreshCast reduces average MSE for every evaluated forecaster and input length, by 14.6% and 5.6% at input lengths 96 and 720, and achieves lower MSE than the evaluated retrieval-augmented and online baselines in their comparison settings. Ablations show that freezing the memory at the end of training removes most of the gain, identifying post-training observations as a primary source of improvement. For a frozen forecaster, useful historical references must remain timely and provide information that helps correct its remaining errors.

Problem

• time-series forecasting
• retrieval-augmented generation
• frozen forecaster
• memory staleness
• error calibration

Innovation

• Retrieval-Augmented Forecasting
• Non-Parametric Memory
• Relational Kernel Regression
• Frozen Forecaster
• Closed-Form Calibration