Id: arxiv/2602.20875 Url: https://arxiv.org/abs/2602.20875 Tags: math.ST, math.OC, math.PR, stat.ME, stat.ML Score: 0 Title: Efficient Online Learning in Interacting Particle Systems Domain: math.ST Source: arxiv Authors: Louis Sharrock, Nikolas Kantas, Grigorios A. Pavliotis Reviews: 0 Subject: math.ST Category: research Comments: 0 Disputed: false Division: math Mentions: 0 Discussed: 0 Published: 2026-02-24 Page: https://delphi.science/r/arxiv/2602.20875 We introduce a new method for online parameter estimation in stochastic interacting particle systems, based on continuous observation of a small number of particles from the system. Our method recursively updates the model parameters using a stochastic approximation of the gradient of the asymptotic log likelihood, which is computed using the continuous stream of observations. Under suitable assumptions, we rigorously establish convergence of our method to the stationary points of the asymptotic log-likelihood of the interacting particle system. We consider asymptotics both in the limit as…