REVIN normalization显著提升时间序列模型零样本泛化能力
A Comparative Study on How Data Normalization Affects Zero-Shot Generalization in Time Series Foundation Models
- 采用REVIN归一化方法，统一处理跨领域数据尺度差异
- 零样本MASE降低89%（相比未归一化基线），优于其他方法44%
- 适用于多种模型架构，尤其对损失敏感度有优化价值
我们研究了时间序列基础模型（TSFMs）的输入归一化方法。尽管归一化在特定数据集的时间序列模型中已有充分研究，但在强调泛化能力的TSFMs中仍被忽视。时间序列数据相较于文本或图像存在显著的域间与通道间尺度差异，并伴随非平稳性，会严重影响TSFM性能，无论其架构多么复杂。通过对四种架构各异的TSFMs进行系统评估，我们实证发现REVIN是效率最高的方法：相较于未归一化基线，零样本MASE降低89%，相较于其他归一化方法降低44%，且无需任何数据集级预处理即可达到最优的域内准确率（0.84 MASE），实现了最佳的准确性-效率权衡。然而其效果依赖于架构设计选择和优化目标，特别是训练损失尺度敏感性和模型类型（概率型、点预测型或基于LLM的模型）。
原文摘要 · Abstract (English)
We investigate input normalization methods for Time-Series Foundation Models (TSFMs). While normalization is well-studied in dataset-specific time-series models, it remains overlooked in TSFMs where generalization is critical. Time-series data, unlike text or images, exhibits significant scale variation across domains and channels, coupled with non-stationarity, can undermine TSFM performance regardless of architectural complexity. Through systematic evaluation across four architecturally diverse TSFMs, we empirically establish REVIN as the most efficient approach, reducing zero-shot MASE by 89\% relative to an un-normalized baseline and by 44\% versus other normalization methods, while matching the best in-domain accuracy (0.84 MASE) without any dataset-level preprocessing -- yielding the highest accuracy-efficiency trade-off. Yet its effect utilization depends on architectural design choices and optimization objective, particularly with respect to training loss scale sensitivity and model type (probabilistic, point-forecast, or LLM-based models).
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。