On Data-Driven Computation of Information Transfer for Causal Inference\n in Dynamical Systems

2018/03/22 by Subhrajit Sinha, Sinha, Subhrajit, Umesh Vaidya +1 · 2 citations
Physics and Astronomy · Neuroscience · #Advanced Thermodynamics and Statistical Mechanics #Model Reduction and Neural Networks #Neural dynamics and brain function

paper · pdf · doi:10.48550/arxiv.1803.08558

Abstract

In this paper, we provide a novel approach to capture causal interaction in a\ndynamical system from time-series data. In citesinha_IT_CDC2016, we have\nshown that the existing measures of information transfer, namely directed\ninformation, granger causality and transfer entropy fail to capture true causal\ninteraction in dynamical system and proposed a new definition of information\ntransfer that captures true causal interaction. The main contribution of this\npaper is to show that the proposed definition of information transfer in\n citesinha_IT_CDC2016 citesinha_IT_ICC can be computed from time-series\ndata. We use transfer operator theoretic framework involving Perron-Frobenius\nand Koopman operators for the data-driven approximation of the system dynamics\nand for the computation of information transfer. Several examples involving\nlinear and nonlinear system dynamics are presented to verify the efficiency of\nthe developed algorithm.\n

Citations

• Investigating Causal Relations by Econometric Models and Cross-spectral Methods

Cited by

• Information Transfer in Dynamical Systems and Optimal Placement of Actuators and Sensors for Control of Non-equilibrium Dynamics
• Information Based Data-Driven Characterization of Stability and Influence in Power Systems

Related

• A unified theory of information transfer and causal relation
• Information flow within stochastic dynamical systems
• A general theory to estimate Information transfer in nonlinear systems
• Causal conditioning and instantaneous coupling in causality graphs
• Quantifying 'causality' in complex systems: Understanding Transfer Entropy