多变量时间序列基础模型中的幻觉检测与缓解：基于扩散方法
Hallucination Detection and Mitigation with Diffusion in Multi-Variate Time-Series Foundation Models
AI总结：
针对多变量时间序列基础模型，提出幻觉定义、检测与缓解方法，利用扩散模型估计幻觉水平，实验显示缓解可降低高达47.7%的幻觉。
摘要：
自然语言处理的基础模型对幻觉有多种连贯的定义，并具备相应的检测与缓解方法。然而，多变量时间序列（MVTS）基础模型尚缺乏类似的定义与方法。我们提出了针对MVTS幻觉的新定义，并利用扩散模型估计幻觉水平，从而开发出新的检测与缓解方法。我们从流行的时序数据集中衍生出关系型数据集，用以基准测试这些关系型幻觉水平。利用这些定义和模型，我们发现开源的预训练MVTS插补基础模型在关系型幻觉上的平均表现，其幻觉程度最高可达弱基线的59.5%。所提出的缓解方法可将这些模型的幻觉水平最多降低47.7%。这些定义和方法有望提升MVTS基础模型的采用率与安全使用。
英文摘要：
Foundation models for natural language processing have many coherent definitions of hallucination and methods for its detection and mitigation. However, analogous definitions and methods do not exist for multi-variate time-series (MVTS) foundation models. We propose new definitions for MVTS hallucination, along with new detection and mitigation methods using a diffusion model to estimate hallucination levels. We derive relational datasets from popular time-series datasets to benchmark these relational hallucination levels. Using these definitions and models, we find that open-source pre-trained MVTS imputation foundation models relationally hallucinate on average up to 59.5% as much as a weak baseline. The proposed mitigation method reduces this by up to 47.7% for these models. The definition and methods may improve adoption and safe usage of MVTS foundation models.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
