Variational Inference of Parameters in Opinion Dynamics Models

Modeling human behavior through the lens of online social networks presents both a significant opportunity and a challenge for understanding complex social phenomena, such as misinformation spread, opinion formation and polarization. While agent-based models (ABMs) are widely used for studying these social phenomena, parameter estimation remains a challenge, often relying on costly simulation-based heuristics. This work uses variational inference to estimate the parameters of an opinion dynamics ABM by transforming the estimation problem into an optimization task that can be solved directly. Our proposal relies on probabilistic generative ABMs (PGABMs): we start by synthesizing a probabilistic generative model from the ABM rules. Then, we transform the inference process into an optimization problem suitable for automatic differentiation. In particular, we use the Gumbel-Softmax reparameterization for categorical agent attributes and Stochastic Variational Inference for parameter estimation. Moreover, we explore the trade-offs of using variational distributions with different complexities: Normal distributions and Normalizing Flows. We validate our method on a bounded confidence model with agent roles (leaders and followers), by estimating both macroscopic (bounded confidence intervals and backfire thresholds) and microscopic (200 categorical agent-level roles) parameters more accurately than simulation-based and MCMC methods.