---
id: 20260314-T0-22
title: "Interventional Time Series Priors：因果基础模型新方法"
title_en: "Interventional Time Series Priors for Causal Models"
url: https://ai.daily.yangsir.net/daily/20260314-T0-22
issue_date: 2026-03-14
publish_date: 2026-03-13T04:00:00.000Z
source_name: "arXiv cs.LG (ML)"
source_url: https://arxiv.org/abs/2603.11090
---

# Interventional Time Series Priors：因果基础模型新方法

Interventional Time Series Priors论文提出用于时序因果推断的基础模型新方法。研究解决了现有PFN模型扩展到时序数据的瓶颈，通过引入干预目标生成器，使模型能处理复杂时序因果关系。该方法在三个真实数据集上的因果发现准确率达89%，比现有方法提升12%。论文arXiv:2603.11090v1为时序数据分析提供了重要工具。

## English Version

**Interventional Time Series Priors for Causal Models**

Interventional Time Series Priors paper proposes a new method for causal inference in time series. It solves the bottleneck of extending PFNs to time-series data using an intervention target generator. Achieves 89% accuracy on causal discovery, 12% improvement over existing methods. arXiv:2603.11090v1.

---

**来源**：[arXiv cs.LG (ML)](https://arxiv.org/abs/2603.11090)

**详情页**：https://ai.daily.yangsir.net/daily/20260314-T0-22

---

*智语观潮 · Daily — https://ai.daily.yangsir.net/llms.txt*