Cluster Prediction for Opinion Dynamics From Partial Observations

We present a Bayesian approach to predict the clustering of opinions for system interacting agents from partial observations. The formulation overcomes unobservability and quantifies uncertainty in prediction. characterize by posterior clusters’ sizes centers, we represent samples. To overcome challenge sampling high-dimensional posterior, introduce an auxiliary implicit (AIS) algorithm using two-step Numerical results show that AIS leads accurate predictions centers leading clusters, both cases noiseless noisy In particular, are predicted with high success rates, but exhibit considerable is sensitive observation noise ratio.