时间序列模型在频谱不匹配时失效，需关注频率特性以提升泛化能力。
Frequency Matters: When Time Series Foundation Models Fail Under Spectral Shift
- 识别频谱偏移是导致时间序列基础模型失效的关键原因。
- 工业级游戏用户行为预测中，模型性能显著低于领域适配基线。
- 提出新训练与评估方法，强调频谱多样性的重要性。
时间序列基础模型（TSFMs）在公开基准上表现优异，被类比为时间序列领域的“BERT时刻”。然而其在工业场景中的有效性仍存疑。本文分析发现，下游任务与预训练阶段主导频率成分不匹配（即频谱偏移）是关键原因。基于移动游戏中的大规模玩家参与度预测任务，我们发现TSFMs表现劣于领域自适应基线。通过设计受控的合成实验，对比已见与未见频段信号，观察到频谱不匹配下性能系统性下降。研究结果表明，频率感知对鲁棒部署至关重要，推动需建立显式考虑频谱多样性的新型预训练与评估协议。
原文摘要 · Abstract (English)
Time series foundation models (TSFMs) have shown strong results on public benchmarks, prompting comparisons to a "BERT moment" for time series. Their effectiveness in industrial settings, however, remains uncertain. We examine why TSFMs often struggle to generalize and highlight spectral shift (a mismatch between the dominant frequency components in downstream tasks and those represented during pretraining) as a key factor. We present evidence from an industrial-scale player engagement prediction task in mobile gaming, where TSFMs underperform domain-adapted baselines. To isolate the mechanism, we design controlled synthetic experiments contrasting signals with seen versus unseen frequency bands, observing systematic degradation under spectral mismatch. These findings position frequency awareness as critical for robust TSFM deployment and motivate new pretraining and evaluation protocols that explicitly account for spectral diversity.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。