AutoPSO automates AI optimization with meta-framework

AutoPSO is a meta-framework that automatically builds customized Particle Swarm Optimization (PSO) algorithms. It uses a bi-level process: an outer search explores a curated pool of PSO components, while an inner loop tests candidate variants on the target task. By leveraging EvoX for population tensorization and batched evaluations, AutoPSO can efficiently assess thousands of particles within practical time budgets. In experiments on numerical benchmarks and neuroevolution robotic control tasks, AutoPSO consistently discovers novel PSO variants that significantly outperform strong baselines. The authors also note that mainstream PSO implementations remain CPU-bound, which limits scalability and increases computational cost. Source code will be made publicly available.