Successful network inference from time-series data using mutual information rate.

Area of Science:

• Complex Systems Analysis
• Information Theory
• Dynamical Systems

Background:

• Inferring connectivity in complex systems from time-series data is crucial for understanding their behavior.
• Traditional methods may struggle with heterogeneous networks or varying data characteristics.

Purpose of the Study:

• To develop and validate an information-based methodology for inferring complex system connectivity.
• To introduce a robust method for estimating Mutual Information Rate (MIR) from noisy, low-resolution time-series data.

Main Methods:

• Analytical derivation of an expression for Mutual Information Rate (MIR).
• Application of normalized MIR for connectivity inference in small networks of dynamical systems.
• Testing the methodology on heterogeneous networks with varying lengths, coupling strengths, and additive noise.

Main Results:

• Successfully inferred connectivity in complex systems using the derived MIR expression.
• Demonstrated robustness to heterogeneous network structures, time-series lengths, coupling strengths, and additive noise.
• Showcased superior performance over standard Mutual Information in networks with differing node time-scale dynamics.

Conclusions:

• The proposed MIR-based methodology offers a powerful tool for network connectivity inference.
• This approach is effective even in challenging conditions where traditional methods fail, such as differing time scales.
• The method advances the analysis of complex dynamical systems.