VCDF: A Validated Consensus-Driven Framework for Time Series Causal Discovery

2026-02-22

AI Summary

Existing methods for causal discovery in time series are often sensitive to noise, non-stationarity, and sampling variability, leading to unstable results. This work proposes VCDF, a consensus-driven framework that is agnostic to the underlying causal discovery algorithm. By integrating temporal block resampling, stability assessment of inferred causal relationships, and consensus fusion, VCDF substantially enhances robustness without modifying the original algorithm. The framework is compatible with mainstream models such as VAR-LiNGAM and PCMCI. Empirical evaluations on synthetic data demonstrate consistent improvements, boosting the F1 score of VAR-LiNGAM by 0.08–0.12 (up to 0.18 for long sequences). Furthermore, experiments on fMRI simulations and IT monitoring scenarios confirm its superior accuracy in recovering true causal structures.

Technology Category

• Machine Learning: Causal Learning
• Reasoning under Uncertainty: Causality
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Responsible Web: Algorithmic accountability and transparency on the web

Abstract

Time series causal discovery is essential for understanding dynamic systems, yet many existing methods remain sensitive to noise, non-stationarity, and sampling variability. We propose the Validated Consensus-Driven Framework (VCDF), a simple and method-agnostic layer that improves robustness by evaluating the stability of causal relations across blocked temporal subsets. VCDF requires no modification to base algorithms and can be applied to methods such as VAR-LiNGAM and PCMCI. Experiments on synthetic datasets show that VCDF improves VAR-LiNGAM by approximately 0.08-0.12 in both window and summary F1 scores across diverse data characteristics, with gains most pronounced for moderate-to-long sequences. The framework also benefits from longer sequences, yielding up to 0.18 absolute improvement on time series of length 1000 and above. Evaluations on simulated fMRI data and IT-monitoring scenarios further demonstrate enhanced stability and structural accuracy under realistic noise conditions. VCDF provides an effective reliability layer for time series causal discovery without altering underlying modeling assumptions.

Problem

• time series causal discovery
• noise sensitivity
• non-stationarity
• sampling variability
• causal stability

Innovation

• causal discovery
• time series
• robustness
• consensus framework
• method-agnostic