GPU-Accelerated PSO for Neural Network-Based Energy Consumption Prediction Capel Tuñón, Manuel Isidoro Salguero Hidalgo, Alberto Gabriel Holgado Terriza, Juan Antonio GPU Particle Swarm Optimization Energy Consumption Optimization Optimization Research Predictive Accuracy This study incorporates advanced parallelism methods using GPU computing to accelerate the process of convergence to an objective, providing faster results for Particle Swarm Optimization (PSO), a bio-inspired stochastic optimization algorithm used to make predictions in various fields. The two proposed distributed implementations of PSO with Apache Spark further enable comprehensive optimisation of both the algorithm structure and its parameters, leading to improved predictive accuracy . Therefore, this approach provides a new and inno vative solution in the field of energy consumption prediction, which can be implemented in a distributed edge–computing solution with optimal performance. 2025-09-08T09:38:30Z 2025-09-08T09:38:30Z 2025-06-05 book part S. Caino-Lores et al. (Eds.): Euro-Par 2024, LNCS 15386, pp. 275–279, 2025 https://hdl.handle.net/10481/106134 10.1007/978-3-031-90203-1_26 eng Lecture Notes in Computer Science;15386 http://creativecommons.org/licenses/by-nd/4.0/ open access Attribution-NoDerivatives 4.0 Internacional Springer Nature Switzerland AG 2025