Newest papers

SCDM: Scalable Causal Discovery in Nonlinear Temporal Systems with Meta-Learning

Jingbo Wang, Kegeng Tang, Zihao Wang, Shaogang Ren · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.29992 · cited by 0 works (OpenAlex) · open-access PDF

Spanish

El descubrimiento de errores causales de las series de tiempo multivariadas no lineales es desafiante en los sistemas de alta dimensión, donde el número de relaciones dirigidas por candidatos crece cuadráticamente con el número de variables. Los métodos existentes a menudo se limitan por el ajuste del modelo orientado hacia el objetivo, las pruebas condicionales repetidas o los procedimientos de búsqueda de metas de gráficos costosos. Proponemos SCDM, un marco de meta-aprendizaje de parámetros compartidos para el descubrimiento de causas temporales escalables en todo el sistema temporal. SCDM trata cada variable objetivo como una tarea mientras aprende un predictor temporal compartido y una matriz de fuerza causal compartida entre tareas. Cada episodio actualiza solo un subconjunto de tareas variables orientadas, lo

Machine translation (Meta M2M100); the English original follows.

Abstract

Causal discovery from nonlinear multivariate time series is challenging in high-dimensional systems, where the number of candidate directed relations grows quadratically with the number of variables. Existing methods are often limited by target-wise model fitting, repeated conditional testing, or expensive graph-search procedures. We propose SCDM, a shared-parameter meta-learning framework for scalable temporal causal discovery. SCDM treats each target variable as a task while learning a shared temporal predictor and a shared causal-strength matrix across tasks. Each meta-episode updates only a subset of target-variable tasks, allowing the model to accumulate structural evidence across the full system without exhaustive target-wise optimization. After training, a post-training graph readout converts the learned predictor into continuous directed causal scores for threshold-independent evaluation. We also provide a finite-error recovery principle showing that, under fully observed delayed-system assumptions and a positive risk-gap condition, thresholding SCDM scores recovers the graph when statistical, meta-optimization, and readout errors are below the population margin. Experiments on controlled synthetic, neuroimaging-inspired, high-dimensional, and real-world temporal benchmarks demonstrate competitive causal recovery and improved scalability in large temporal systems.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Scalability, Directed acyclic graph, Thresholding, Artificial intelligence, Machine learning

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange