NIPS: Network Inference with Partial State measurements using forced-delay embedding

Area of Science:

• Complex Systems
• Network Science
• Dynamical Systems

Background:

• Understanding complex network dynamics requires decoding connectivity patterns from time series data.
• Existing network inference methods often require complete state measurements, which are impractical in real-world scenarios.

Purpose of the Study:

• To introduce a novel framework, Network Inference from Partial States (NIPS), for reconstructing network structures from partial-state observations.
• To enable accurate network inference when full-state measurements are unavailable.

Main Methods:

• Developed NIPS by modeling coupling inputs as external forcing and applying forced-delay embedding theory.
• Established a map linking node observable evolution to observable state components, focusing on self-dependent dynamics.

Main Results:

• Demonstrated accurate network reconstruction using simulated and experimental data, even with limited observations.
• Evaluated the robustness of NIPS against noisy data and hidden network nodes.

Conclusions:

• NIPS provides a robust method for network reconstruction from partial data, overcoming limitations of existing approaches.
• The framework was successfully extended to handle networks coupled through unobservable states, broadening its applicability.