Time Series Causal Discovery Based on Data Augmentation and Causal Validation Hefei University of Technology, School of Computer Science a.

Format:

Book

Conference/Event

Author/Creator:

Yang, Jing, author.

Chen, Xiaotao, author.

Qin, Xuanli, author.

Xu, Xianjun, author.

Hu, Zhangxiang, author.

Conference Name:

Interntional Conference on the New Energy and Intelligent Vehicles (2025-11-02 : Hefei, China)

Language:

English

Subjects (All):

Machine learning.

Noise.

Research and development.

Tools and equipment.

Weather and climate.

Local Subjects:

Machine learning.

Noise.

Research and development.

Tools and equipment.

Weather and climate.

Physical Description:

1 online resource

Place of Publication:

Warrendale, PA SAE International 2026

Summary:

Causal discovery within time series is crucial for revealing the actual causal mechanisms in dynamic systems, and it has major impacts in various fields like economics, healthcare, and climate science. Even though it's important, accurately figuring out causal relationships from observational temporal data is still quite a difficult task. Traditional Granger causality based methods are often limited by noise sensitivity, large amount of data, and the inability to distinguish between real causality and false correlation caused by hidden factors.In order to solve these problems, this paper presents CausalAugVeri, which is a new algorithm that cleverly mixes data augmentation with causal verification to make causal discovery more solid and precise. This work has three main points: First, we carefully check that using convolutional data augmentation techniques can greatly improve how well time series predictions work, giving a steadier base for detecting Granger causality. Second, the suggested method rebuilds cause variables using specific intervention ways and adds a causal verification part that strictly removes wrong findings, keeping only real causal connections. Third, we do thorough experiments on both made-up and real-world time series datasets, showing that CausalAugVeri always does better than current top methods, particularly when there's little data and lots of noise.The results prove that our way gives a dependable and expandable answer for causal discovery in complicated time-related situations, connecting the gap between augmentation based on deep learning and traditional causality analysis. This study not only gives a methodologically strong structure but also offers practical tools for real-world uses that need sturdy causal understanding

Notes:

Vendor supplied data

Access Restriction:

Restricted for use by site license