2021/05/12 by Wagner Gonçalves Pinto, Pinto, Wagner Gonçalves, Antonio Alguacil +3 · 1 citation
Computer Science · Earth and Planetary Sciences · #FOS: Computer and information sciences #FOS: Physical sciences #Fluid Dynamics (physics.flu-dyn) #Image and Signal Denoising Methods #Machine Learning (cs.LG) #Meteorological Phenomena and Simulations #Underwater Acoustics Research
paper · pdf · doi:10.48550/arxiv.2105.05482
openalex publication_date 2021/05/12 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Reproducibility of a deep-learning fully convolutional neural network is evaluated by training several times the same network on identical conditions (database, hyperparameters, hardware) with non-deterministic Graphics Processings Unit (GPU) operations. The propagation of two-dimensional acoustic waves, typical of time-space evolving physical systems, is studied on both recursive and non-recursive tasks. Significant changes in models properties (weights, featured fields) are observed. When tested on various propagation benchmarks, these models systematically returned estimations with a high level of deviation, especially for the recurrent analysis which strongly amplifies variability due to the non-determinism. Trainings performed with double floating-point precision provide slightly better estimations and a significant reduction of the variability of both the network parameters and its testing error range.