Skip to main content
U.S. Department of Energy
Office of Scientific and Technical Information

In situ compression artifact removal in scientific data using deep transfer learning and experience replay

Journal Article · · Machine Learning: Science and Technology
The massive amount of data produced during simulation on high-performance computers has grown exponentially over the past decade, exacerbating the need for streaming compression and decompression methods for efficient storage and transfer of this data---key to realizing the full potential of large-scale computational science. Lossy compression approaches such as JPEG when applied to scientific simulation data realized as a stream of images can achieve good compression rates but at the cost of introducing compression artifacts and loss of information. This paper develops a unified framework for in situ compression artifact removal in which the fully convolutional neural network architectures are combined with scalable training, transfer learning, and experience replay to achieve superior accuracy and efficiency while significantly decreasing the storage footprint as compared with the traditional optimization-based approaches. We demonstrate the proposed approach and compare it with compressed sensing postprocessing and other baseline deep learning models using climate simulations and nuclear reactor simulations, both of which are driven by hyperbolic partial differential equations. Our approach when applied to remove the compression artifacts on the JPEG-compressed nuclear reactor simulation data (using a transfer-trained model that was pretrained on the climate simulation data and updated incrementally as the nuclear reactor simulation progressed), achieved a significant improvement---mean peak signal-to-noise ratio of 42.438 as compared with 27.725 obtained with the compressed sensing approach.
Sponsoring Organization:
USDOE Office of Science (SC), Advanced Scientific Computing Research (ASCR)
Grant/Contract Number:
SC0012704; AC52-07NA27344; AC02-06CH11357; AC05-00OR22725
OSTI ID:
1755287
Alternate ID(s):
OSTI ID: 1814369
OSTI ID: 1755288
OSTI ID: 1693391
Journal Information:
Machine Learning: Science and Technology, Journal Name: Machine Learning: Science and Technology Journal Issue: 2 Vol. 2; ISSN 2632-2153
Publisher:
IOP PublishingCopyright Statement
Country of Publication:
United Kingdom
Language:
English

References (40)

Stable signal recovery from incomplete and inaccurate measurements
  • Candès, Emmanuel J.; Romberg, Justin K.; Tao, Terence
  • Communications on Pure and Applied Mathematics, Vol. 59, Issue 8, p. 1207-1223 https://doi.org/10.1002/cpa.20124
journal January 2006
Enhancing Sparsity by Reweighted ℓ 1 Minimization journal October 2008
Image Reconstruction from Fourier Data Using Sparsity of Edges journal December 2014
Image Reconstruction from Undersampled Fourier Data Using the Polynomial Annihilation Transform journal September 2015
Nonlinear total variation based noise removal algorithms journal November 1992
A standard test set for numerical approximations to the shallow water equations in spherical geometry journal September 1992
A new spherical harmonics scheme for multi-dimensional radiation transport I. Static matter configurations journal June 2013
DGM: A deep learning algorithm for solving partial differential equations journal December 2018
Deep visual domain adaptation: A survey journal October 2018
Numerical integration on the sphere journal January 1982
Stochastic finite element methods for partial differential equations with random input data journal May 2014
Solving high-dimensional partial differential equations using deep learning journal August 2018
A Comparison of Moment Closures for Linear Kinetic Transport Equations: The Line Source Benchmark journal September 2013
Catastrophic Forgetting, Rehearsal and Pseudorehearsal journal June 1995
Polynomial approximation via compressed sensing of high-dimensional functions on lower sets journal September 2017
An optimization approach for removing blocking effects in transform coding journal April 1995
Gradient Projection for Sparse Reconstruction: Application to Compressed Sensing and Other Inverse Problems journal December 2007
An Interior-Point Method for Large-Scale -Regularized Least Squares journal December 2007
Optimizing Scientist Time through In Situ Visualization and Analysis journal January 2018
Discrete Cosine Transform journal January 1974
Enhancing Quality for HEVC Compressed Videos journal July 2019
Compression Artifact Reduction by Overlapped-Block Transform Coefficient Estimation With Block Similarity journal December 2013
Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising journal July 2017
Robust uncertainty principles: exact signal reconstruction from highly incomplete frequency information journal February 2006
Compressed sensing journal April 2006
Near-Optimal Signal Recovery From Random Projections: Universal Encoding Strategies? journal January 2006
A Survey on Transfer Learning journal October 2010
Reducing Artifacts in JPEG Decompression Via a Learned Dictionary journal February 2014
Signal Recovery by Proximal Forward-Backward Splitting journal January 2005
Bregman Iterative Algorithms for $\ell_1$-Minimization with Applications to Compressed Sensing journal January 2008
The Split Bregman Method for L1-Regularized Problems journal January 2009
NESTA: A Fast and Accurate First-Order Method for Sparse Recovery journal January 2011
Augmented Lagrangian Method, Dual Methods, and Split Bregman Iteration for ROF, Vectorial TV, and High Order Models journal January 2010
High-Order Entropy-Based Closures for Linear Transport in Slab Geometry II: A Computational Study of the Optimization Problem journal January 2012
waveSZ: a hardware-algorithm co-design of efficient lossy compression for scientific data
  • Tian, Jiannan; Di, Sheng; Zhang, Chengming
  • PPoPP '20: 25th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming, Proceedings of the 25th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming https://doi.org/10.1145/3332466.3374525
conference February 2020
Moving Vortices on the Sphere: A Test Case for Horizontal Advection Problems journal February 2008
A Discontinuous Galerkin Global Shallow Water Model journal April 2005
An initial-value problem for testing numerical models of the global shallow-water equations journal January 2004
Convergence of filtered spherical harmonic equations for radiation transport journal January 2016
Rudin-Osher-Fatemi Total Variation Denoising using Split Bregman journal January 2012

Similar Records

Related Subjects