MetaCaster: Meta-Harness-Optimized Agent for End-to-End Few-Shot Learning of Lightweight Time Series Forecasters

2026-08-24

AI Summary

研究针对资源受限场景下轻量级时间序列预测器的少样本学习问题，提出MetaCaster框架，通过多代理数据生成自动训练专用预测模型。

Abstract

Time series forecasting (TSF) is evolving toward multimodal and agentic settings, yet using foundation models remains uneconomical in resource-constrained scenarios, where compact, specialized forecasters are more desirable. However, lightweight forecasters typically require substantial training data, limiting their use in domains with scarce, slowly accumulated, or privacy-sensitive time series. To address this dilemma, we investigate the challenging problem of few-shot learning for lightweight forecasters. We propose MetaCaster, a meta-harness-optimized multi-agent framework that uses agentic data generation to automatically train specialized lightweight forecasters from only a few examples and textual contexts. Our work highlights a new TSF paradigm in which agents act not as forecasters but as intermediary engineers that prepare efficient, task-specific forecasters for deployment. Experiments on 18 datasets, 23 state-of-the-art lightweight forecasters, and 14 baselines demonstrate that MetaCaster achieves both data efficiency and computational efficiency while maintaining high-quality TSF performance.

Problem

• Time Series Forecasting
• Few-Shot Learning
• Lightweight Forecasters

Innovation

• few-shot learning
• lightweight forecasters
• meta-harness-optimized
• agentic data generation
• time series forecasting