Castor: Causal Temporal Regime Structure Learning

Abdellah Rahmani, Pascal Frossard

Primary Area: causal reasoning

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Keywords: Causal structure learning, time series, heterogeneous time series, Causal discovery

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.

Abstract:

The task of uncovering causal relationships among variables from time series data stands as an essential and challenging objective that cuts across a broad array of disciplines ranging from climate science to healthcare. Time series entail linear or non-linear relationships, and usually follow multiple a priori unknown regimes. Existing causal discovery methods can infer summary causal graphs from heterogeneous data with known regime indices, but they fall short in comprehensively learning both regime indices and the full temporal causal graph.
In this paper, we introduce CASTOR, a novel framework designed to learn causal relationships in heterogeneous time series data composed of various regimes, each governed by a distinct causal graph. Through the maximization of a score function via the EM algorithm, CASTOR infers the number of regimes and learns linear or non-linear causal relationships inherent in each regime. We demonstrate the robust convergence properties of CASTOR, specifically highlighting its proficiency in accurately identifying unique regimes. Empirical evidence, garnered from exhaustive synthetic experiments and two real-world benchmarks, confirm CASTOR's superior performance in causal discovery compared to relevant baselines.
By learning a full temporal causal graph for each regime, CASTOR establishes itself as a distinctly interpretable method for causal discovery in heterogeneous time series.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 6010