The appropriateness of ignorance in the inverse kinetic Ising model
arXiv:1612.06185 · doi:10.1088/1751-8121/aa59dc
Abstract
We develop efficient ways to consider and correct for the effects of hidden units for the paradigmatic case of the inverse kinetic Ising model with fully asymmetric couplings. We identify two sources of error in reconstructing the connectivity among the observed units while ignoring part of the network. One leads to a systematic bias in the inferred parameters, whereas the other involves correlations between the visible and hidden populations and has a magnitude that depends on the coupling strength. We estimate these two terms using a mean field approach and derive self-consistent equations for the couplings accounting for the systematic bias. Through application of these methods on simple networks of varying relative population size and connectivity strength, we assess how and under what conditions the hidden portion can influence inference and to what degree it can be crudely estimated. We find that for weak to moderately coupled systems, the effects of the hidden units is a simple rotation that can be easily corrected for. For strongly coupled systems, the non-systematic term becomes large and can no longer be safely ignored, further highlighting the importance of understanding the average strength of couplings for a given system of interest.
References in corpus (15)
- Weak pairwise correlations imply strongly correlated network states in a neural population
- The Ising Model for Neural Data: Model Quality and Approximate Methods for Extracting Functional Connectivity
- Crises and collective socio-economic phenomena: simple models and challenges
- Mean Field Theory For Non-Equilibrium Network Reconstruction
- Correlations and functional connections in a population of grid cells
- Learning and inference in a nonequilibrium Ising model with hidden nodes
- U.S. stock market interaction network as learned by the Boltzmann Machine
- Network Inference with Hidden Units
- Belief-Propagation and replicas for inference and learning in a kinetic Ising model with hidden spins
- Inferring hidden states in a random kinetic Ising model: replica analysis
- Statistical pairwise interaction model of stock market
- Effect of coupling asymmetry on mean-field solutions of direct and inverse Sherrington-Kirkpatrick model
- Generalized mean field approximation for parallel dynamics of the Ising model
- Effects of hidden nodes on network structure inference
- Financial interaction networks inferred from traded volumes