On the reproducibility of fully convolutional neural networks for modeling time-space evolving physical systems
Not on the map. Why?
Abstract (arXiv)
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.
Key facts. Cited by 3 papers in the map (data through 7 Oct 2026). Builds on 22 of its 53 references that are in the map. First posted on arXiv 12 May 2021.
Builds on
22 of 53 references.
References · 22
- 2014Adam: A Method for Stochastic Optimizationcited by 36,960
- 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,160
- 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,604
- 2015Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classificationcited by 3,590
- 2017Mixed Precision Trainingcited by 1,135
- 2017Visualizing the Loss Landscape of Neural Netscited by 1,126
- 2017DGM: A deep learning algorithm for solving partial differential equationscited by 674
- 2019DeepXDE: A deep learning library for solving differential equationscited by 617
- 2019Machine Learning for Fluid Mechanicscited by 380
- 2020Improving Reproducibility in Machine Learning Research (A Report from the NeurIPS 2019 Reproducibility Program)cited by 318
- 2015Deep multi-scale video prediction beyond mean square errorcited by 297
- 2019Random Search and Reproducibility for Neural Architecture Searchcited by 253
10 more
- 2015Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networkscited by 214
- 2020Solver-in-the-Loop: Learning from Differentiable Physics to Interact with Iterative PDE-Solverscited by 201
- 2016Accelerating Eulerian Fluid Simulation With Convolutional Networkscited by 128
- 2017A unified deep artificial neural network approach to partial differential equations in complex geometriescited by 122
- 2019PPINN: Parareal Physics-Informed Neural Network for time-dependent PDEscited by 116
- 2019A Step Toward Quantifying Independently Reproducible Machine Learning Researchcited by 77
- 2019Poisson CNN: Convolutional neural networks for the solution of the Poisson equation on a Cartesian meshcited by 22
- 2020Comparing recurrent and convolutional neural networks for predicting wave propagationcited by 15
- 2018Cyberbullying Detection in Social Networks Using Deep Learning Based Models; A Reproducibility Studycited by 14
- 2018Approximating the solution to wave propagation using deep neural networkscited by 8
Built on by
Citing papers per year: 2022: 1; 2023: 0; 2024: 0; 2025: 1; 2026: 1.
3 papers.