Abstract:The inference performance of the pseudolikelihood method is discussed in the framework of the inverse Ising problem when the l (2)-regularized (ridge) linear regression is adopted. This setup is introduced for theoretically investigating the situation where the data generation model is different from the inference one, namely the model mismatch situation. In the teacher-student scenario under the assumption that the teacher couplings are sparse, the analysis is conducted using the replica and cavity methods, with a special focus on whether the presence/absence of teacher couplings is correctly inferred or not. The result indicates that despite the model mismatch, one can perfectly identify the network structure using naive linear regression without regularization when the number of spins N is smaller than the dataset size M, in the thermodynamic limit N -> infinity. Further, to access the underdetermined region M < N, we examine the effect of the l (2) regularization, and find that biases appear in all the coupling estimates, preventing the perfect identification of the network structure. We, however, find that the biases are shown to decay exponentially fast as the distance from the center spin chosen in the pseudolikelihood method grows. Based on this finding, we propose a two-stage estimator: in the first stage, the ridge regression is used and the estimates are pruned by a relatively small threshold; in the second stage the naive linear regression is conducted only on the remaining couplings, and the resultant estimates are again pruned by another relatively large threshold. This estimator with the appropriate regularization coefficient and thresholds is shown to achieve the perfect identification of the network structure even in 0 < M/N < 1. Results of extensive numerical experiments support these findings.

Data-informativity conditions for structured linear systems with implications for dynamic networks

Structure Learning in Inverse Ising Problems Using ℓ_2-Regularized Linear Estimator

Identifiability and Identification of Switching Dynamical Networks: A Data-Based Approach

Informative Conditions for a Data Set in an MIMO Networked Control System

Single module identifiability in linear dynamic networks

A Consequence of Invertibility on the Normal Form of a MIMO Nonlinear Systems

Depicting Network Structures from Variable Data Produced by Unknown Colored-Noise Driven Dynamics

Stabilization by Output Feedback of Multivariable Invertible Nonlinear Systems.

The static and dynamic analysis of network data using information theory

Structure Reconstruction for Linear Network Systems with Dynamical Structure Functions

Informative Conditions for the Data Set in an MIMO Networked Control System with Delays, Packet Dropout and Transmission Scheduling

Model identification of a network as compressing sensing

Identifiability in Dynamic Acyclic Networks with Partial Excitations and Measurements

Data-Driven Identification of Networks of Dynamic Systems

Network Identifiability from Intrinsic Noise.

On Identification of Dynamical Structure Functions: A Sparse Bayesian Learning Approach.

Dynamical Structure Functions for the Estimation of LTI Networks with Limited Information

Strong Structural Controllability of Structured Networks with MIMO node systems

Structured stability analysis of networked systems with uncertain links

Inferring biological networks by sparse identification of nonlinear dynamics

Informative Conditions for a Data Set in a MIMO Closed‐loop System with a Switching Controller