A causal discovery method named Intervened TaylorKAN (ITK), which simulates controlled experiments within reconstructed systems to provide more intuitive causal inferences and achieves high accuracy on synthetic data with known causal structures, confirming its validity.
Abstract
Identifying causal relationships from observational time-series data is a central challenge in understanding complex dynamical systems. However, few existing methods attempt to capture the intuitive human understanding of causality: that an intervention on one variable leads to a subsequent change in another. To bridge this gap, this paper introduces a causal discovery method named Intervened TaylorKAN (ITK), which simulates controlled experiments within reconstructed systems. Specifically, we leverage the recently emerging Kolmogorov-Arnold Networks (KANs) with Taylor series as activation functions to construct a surrogate model that faithfully replicates the original system's dynamics. The core of ITK involves applying targeted interventions on this reconstructed system and observing the responses in other variables, thereby providing more intuitive causal inferences. We evaluated ITK's performance through extensive experiments on both synthetic and real-world datasets. Results demonstrate that ITK achieves high accuracy on synthetic data with known causal structures, confirming its validity. When applied to real-world data, ITK produces reasonable and interpretable causal conclusions, highlighting its potential to decipher causal relationships in complex natural systems.
Time series causal discovery --- the process of detecting causation between time series --- has its roots in signal processing research from over a half century ago. It has since received active and widespread interest, evolving into different forms across domains such as statistics, dynamical systems and machine learn...
Kurt Butler, Damian Machlanski, Michael P. J. Camilleri et al.· 0 citations
Unveiling directional causality in complex, high-dimensional dynamical systems remains a fundamental challenge. Conventional nonparametric approaches, such as Convergent Cross Mapping, require computationally intensive surrogate tests and frequently yield false-positive detections under strong coupling due to generaliz...
Structural causal models for time series recover contemporaneous and lagged effects, but most methods require complete observation windows and become misspecified when samples are missing. We introduce ITSY, the first continuous-optimization method for causal discovery from irregular time series under a linear model. I...
Wen-Bo Xu, Yue He, Yun-Hai Wang et al.· arXiv.org· 0 citations
While statistical learning has made remarkable progress in recent decades, driven by large datasets and data-driven protocols that have transformed scientific practice, it still faces a key limitation: it struggles to generalize beyond the data it has seen, often failing under distribution shift. Causality offers a pro...
Causal discovery from nonstationary subsample time series is a fundamental yet challenging problem in many scientific domains. In such data, observations are often recorded at a coarse time scale (subsampling). At the same time, the underlying data-generating mechanisms may change over time (nonstationarity), making st...
Wei Chen, Qian Huang, Fu-Hong Wang et al.· Neural Networks· 0 citations
Discovering causal structures in multivariate time series (MTS) is critical in domains such as finance and neuroscience, yet remains challenging in practice due to three distinct obstacles. First, MTS are routinely corrupted by missing values, where naive impute-then-discover pipelines distort the underlying data distr...
Abdellah Rahmani, Pascal Frossard· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.