Decoupling approximation robustly reconstructs directed dynamical networks
arXiv:1712.03100 · doi:10.1088/1367-2630/aae941
Abstract
Methods for reconstructing the topology of complex networks from time-resolved observations of node dynamics are gaining relevance across scientific disciplines. Of biggest practical interest are methods that make no assumptions about properties of the dynamics, and can cope with noisy, short and incomplete trajectories. Ideal reconstruction in such scenario requires and exhaustive approach of simulating the dynamics for all possible network configurations and matching the simulated against the actual trajectories, which of course is computationally too costly for any realistic application. Relying on insights from equation discovery and machine learning, we here introduce \textit{decoupling approximation} of dynamical networks and propose a new reconstruction method based on it. Decoupling approximation consists of matching the simulated against the actual trajectories for each node individually rather than for the entire network at once. Despite drastic reduction of the computational cost that this approximation entails, we find our method's performance to be very close to that of the ideal method. In particular, we not only make no assumptions about properties of the trajectories, but provide strong evidence that our methods' performance is largely independent of the dynamical regime at hand. Of crucial relevance for practical applications, we also find our method to be extremely robust to both length and resolution of the trajectories and relatively insensitive to noise.
References in corpus (28)
- Discovering governing equations from data: Sparse identification of nonlinear dynamical systems
- Epidemic processes in complex networks
- Synchronization in complex networks
- The structure and dynamics of multilayer networks
- Characterization of complex networks: A survey of measurements
- Machine learning phases of matter
- Synchronization reveals topological scales in complex networks
- Missing and spurious interactions and the reconstruction of complex networks
- Learning phase transitions by confusion
- Finding Density Functionals with Machine Learning
- Representation of compounds for machine-learning prediction of physical properties
- Coupling functions: Universal insights into dynamical interaction mechanisms
- Data Based Identification and Prediction of Nonlinear and Complex Dynamical Systems
- Quantum Loop Topography for Machine Learning
- Parsimonious module inference in large networks
- Machine Learning for Precise Quantum Measurement
- Revealing networks from dynamics: an introduction
- Robust Reconstruction of Complex Networks from Sparse Data
- Combining complex networks and data mining: why and how
- Inference of Time-Evolving Coupled Dynamical Systems in the Presence of Noise
- Mean Field Theory For Non-Equilibrium Network Reconstruction
- Network reconstruction from random phase-resetting
- Revealing physical interaction networks from statistics of collective dynamics
- Identifying polymer states by machine learning
- Multi-scale structure and topological anomaly detection via a new network statistic: The onion decomposition
- Reconstructing networks of pulse-coupled oscillators from spike trains
- Reconstruction of chaotic neural network from observed firing rates
- Extracting topological features from dynamical measures in networks of Kuramoto oscillators