UnCLe: Towards Scalable Dynamic Causal Discovery in Non-linear Temporal Systems

2025-11-05

AI Summary

Accurately modeling dynamic causal relationships in time-varying nonlinear systems remains challenging due to their evolving, nonlinear interdependencies. Method: This paper proposes a joint framework integrating a decoupling-reconstruction network with an autoregressive dynamic dependency matrix, enabling the first data-point-level, time-resolved estimation of causal influence. It decouples input sequences, incorporates prediction error analysis under temporal perturbations, and explicitly models time-evolving, nonlinear variable dependencies—thereby transcending the limitations of static causal discovery paradigms. Contribution/Results: Evaluated on synthetic and real human motion datasets, the method significantly outperforms existing baselines in causal graph accuracy, time-varying pattern identification, and scalability. It provides a novel, interpretable tool for mechanistic analysis of complex dynamic systems.

Technology Category

• Machine Learning: Causal Learning
• Reasoning under Uncertainty: Causality
• Knowledge Representation and Reasoning: Action, Change, and Causality

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Web Mining and Content Analysis: Models for Web evolution

Abstract

Uncovering cause-effect relationships from observational time series is fundamental to understanding complex systems. While many methods infer static causal graphs, real-world systems often exhibit dynamic causality-where relationships evolve over time. Accurately capturing these temporal dynamics requires time-resolved causal graphs. We propose UnCLe, a novel deep learning method for scalable dynamic causal discovery. UnCLe employs a pair of Uncoupler and Recoupler networks to disentangle input time series into semantic representations and learns inter-variable dependencies via auto-regressive Dependency Matrices. It estimates dynamic causal influences by analyzing datapoint-wise prediction errors induced by temporal perturbations. Extensive experiments demonstrate that UnCLe not only outperforms state-of-the-art baselines on static causal discovery benchmarks but, more importantly, exhibits a unique capability to accurately capture and represent evolving temporal causality in both synthetic and real-world dynamic systems (e.g., human motion). UnCLe offers a promising approach for revealing the underlying, time-varying mechanisms of complex phenomena.

Problem

• Uncovering evolving cause-effect relationships in observational time series data
• Capturing dynamic temporal causality where relationships change over time
• Estimating time-resolved causal graphs for complex real-world systems

Innovation

• Uncoupler and Recoupler networks disentangle semantic representations
• Auto-regressive Dependency Matrices learn inter-variable dependencies
• Dynamic causal influences estimated via temporal perturbation errors