Problem
Research questions and friction points this paper is trying to address.
Identifies causal structure from temporal data
Addresses unknown confounding in time-series analysis
Handles correlated causes and cyclic relationships
Doubly Robust Structure Identification from Temporal Data
🤖 AI Summary
Temporal causal discovery faces challenges including unknown confounders, high-dimensional and strongly correlated latent causes, and potential cyclic dependencies. Method: We propose the first doubly robust temporal structure learning framework, integrating doubly robust estimation, temporal causal discovery, and vector autoregressive modeling. Theoretically, it guarantees asymptotic recovery of the true causal structure under confounded and cyclic causal graphs, and supports nonlinear extensions and noise-robust inference. Contribution/Results: Extensive experiments on synthetic and real-world datasets demonstrate substantial improvements over state-of-the-art baselines: structural identification accuracy increases by over 35% in scenarios with strong confounding, limited samples, and cyclic structures. Our method establishes a new paradigm for high-dimensional temporal causal modeling—uniquely combining rigorous theoretical guarantees with strong empirical performance.
📝 Abstract
Learning the causes of time-series data is a fundamental task in many applications, spanning from finance to earth sciences or bio-medical applications. Common approaches for this task are based on vector auto-regression, and they do not take into account unknown confounding between potential causes. However, in settings with many potential causes and noisy data, these approaches may be substantially biased. Furthermore, potential causes may be correlated in practical applications. Moreover, existing algorithms often do not work with cyclic data. To address these challenges, we propose a new doubly robust method for Structure Identification from Temporal Data ( SITD ). We provide theoretical guarantees, showing that our method asymptotically recovers the true underlying causal structure. Our analysis extends to cases where the potential causes have cycles and they may be confounded. We further perform extensive experiments to showcase the superior performance of our method.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Uses double machine learning for structure identification
Handles cycles and confounding in causal variables
Provides theoretical guarantees for true structure recovery
🔎 Similar Papers
No similar papers found.
