English

Scalable Policy Maximization Under Network Interference

Machine Learning 2026-05-07 v2 Machine Learning

Abstract

Many interventions, such as vaccines in clinical trials or coupons in online marketplaces, must be assigned sequentially without full knowledge of their effects. Multi-armed bandit algorithms have proven successful in such settings. However, standard independence assumptions fail when the treatment status of one individual impacts the outcomes of others, a phenomenon known as interference. We study optimal-policy learning under interference on a dynamic network. Existing approaches to this problem require repeated observations of the same fixed network and struggle to scale in sample size beyond as few as fifteen connected units -- both limit applications. We show that under common assumptions on the structure of interference, rewards become linear. This enables us to develop a scalable Thompson sampling algorithm that maximizes policy impact when a new nn-node network is observed each round. We prove a Bayesian regret bound that is sublinear in nn and the number of rounds. Simulation experiments show that our algorithm learns quickly and outperforms existing methods. The results close a key scalability gap between causal inference methods for interference and practical bandit algorithms, enabling policy optimization in large-scale networked systems.

Keywords

Cite

@article{arxiv.2505.18118,
  title  = {Scalable Policy Maximization Under Network Interference},
  author = {Aidan Gleich and Eric Laber and Alexander Volfovsky},
  journal= {arXiv preprint arXiv:2505.18118},
  year   = {2026}
}
R2 v1 2026-07-01T02:34:21.488Z