arXiv:2410.12326cs.LG2024-10KDD被引 13

LLM在时序任务中表现不佳,因语言与时序数据难以对齐。

Understanding Why Large Language Models Can Be Ineffective in Time Series Analysis: The Impact of Modality Alignment

  • 用线性模型和随机初始化LLM对比,验证其性能优势有限。
  • 简单模型参数少却更稳定,且不破坏时序结构。
  • 现有重编程方法仅实现伪对齐,无法真正融合时序特征。

大型语言模型(LLMs)在时序分析中展现出令人印象深刻的性能,似乎比传统Transformer方法更好地理解时间关系。然而,由于LLMs并非专为时序任务设计,简单的线性回归模型往往能以更低的复杂度达到相当的性能。本研究通过大量实验评估了将LLMs应用于预测、分类、缺失值填补和异常检测等核心时序任务的有效性。将LLMs与单层线性模型及随机初始化的LLMs进行对比,结果表明,在这些任务上LLMs的优势微乎其微,甚至可能扭曲数据的时间结构。相反,简单模型始终表现更优,且所需参数显著更少。此外,我们分析了现有的重编程技术,通过数据流形分析发现,这些方法未能有效将时序数据与语言空间对齐,反而在嵌入空间中表现出“伪对齐”行为。研究结果表明,基于LLM的方法在时序任务中的表现,源于时序数据自身的内在特性与结构,而非与语言模型架构之间的有意义对齐。

原文摘要 · Abstract (English)

Large Language Models (LLMs) have demonstrated impressive performance in time series analysis and seems to understand the time temporal relationship well than traditional transformer-based approaches. However, since LLMs are not designed for time series tasks, simpler models like linear regressions can often achieve comparable performance with far less complexity. In this study, we perform extensive experiments to assess the effectiveness of applying LLMs to key time series tasks, including forecasting, classification, imputation, and anomaly detection. We compare the performance of LLMs against simpler baseline models, such as single layer linear models and randomly initialized LLMs. Our results reveal that LLMs offer minimal advantages for these core time series tasks and may even distort the temporal structure of the data. In contrast, simpler models consistently outperform LLMs while requiring far fewer parameters. Furthermore, we analyze existing reprogramming techniques and show, through data manifold analysis, that these methods fail to effectively align time series data with language and display "pseudo-alignment" behavior in embedding space. Our findings suggest that the performance of LLM based methods in time series tasks arises from the intrinsic characteristics and structure of time series data, rather than any meaningful alignment with the language model architecture.

时序分析LLM局限模态对齐

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。