Potential Flow Generator With L2 Optimal Transport Regularity for Generative Models

Yang, Liu; Karniadakis, George Em

doi:10.1109/tnnls.2020.3028042

Potential Flow Generator With L₂ Optimal Transport Regularity for Generative Models

Journal Article · Tue Oct 20 00:00:00 EDT 2020 · IEEE Transactions on Neural Networks and Learning Systems

DOI:https://doi.org/10.1109/tnnls.2020.3028042· OSTI ID:2281642

^[1]; ^[2]

Brown Univ., Providence, RI (United States); Brown University
Brown Univ., Providence, RI (United States)

Here, we propose a potential flow generator with $$L_{2}$$ optimal transport regularity, which can be easily integrated into a wide range of generative models, including different versions of generative adversarial networks (GANs) and normalizing flow models. With only a slight augmentation to the original generator loss functions, our generator not only tries to transport the input distribution to the target one but also aims to find the one with minimum $$L_{2}$$ transport cost. We show the effectiveness of our method in several 2-D problems and illustrate the concept of “proximity” due to the $$L_{2}$$ optimal transport regularity. Subsequently, we demonstrate the effectiveness of the potential flow generator in image translation tasks with unpaired training data from the MNIST data set and the CelebA data set with a comparison against vanilla Wasserstein GAN with gradient penalty (WGAN-GP) and CycleGAN.

View Accepted Manuscript (DOE)

Research Organization:: Brown Univ., Providence, RI (United States)

Sponsoring Organization:: USDOE

Grant/Contract Number:: SC0019453

OSTI ID:: 2281642

Alternate ID(s):: OSTI ID: 1980528

Journal Information:: IEEE Transactions on Neural Networks and Learning Systems, Journal Name: IEEE Transactions on Neural Networks and Learning Systems Journal Issue: 2 Vol. 33; ISSN 2162-237X

Publisher:: IEEE Computational Intelligence SocietyCopyright Statement

Country of Publication:: United States

Language:: English

References (15)

Polar factorization and monotone rearrangement of vector-valued functions Brenier, Yann Communications on Pure and Applied Mathematics, Vol. 44, Issue 4 https://doi.org/10.1002/cpa.3160440402	journal	June 1991
The geometry of optimal transportation Gangbo, Wilfrid; McCann, Robert J. Acta Mathematica, Vol. 177, Issue 2 https://doi.org/10.1007/BF02392620	journal	January 1996
A computational fluid mechanics solution to the Monge-Kantorovich mass transfer problem Benamou, Jean-David; Brenier, Yann Numerische Mathematik, Vol. 84, Issue 3 https://doi.org/10.1007/s002110050002	journal	January 2000
{Euclidean, metric, and Wasserstein} gradient flows: an overview Santambrogio, Filippo Bulletin of Mathematical Sciences, Vol. 7, Issue 1 https://doi.org/10.1007/s13373-017-0101-1	journal	March 2017
Natural gradient via optimal transport Li, Wuchen; Montúfar, Guido Information Geometry, Vol. 1, Issue 2 https://doi.org/10.1007/s41884-018-0015-3	journal	November 2018
DGM: A deep learning algorithm for solving partial differential equations Sirignano, Justin; Spiliopoulos, Konstantinos Journal of Computational Physics, Vol. 375 https://doi.org/10.1016/j.jcp.2018.08.029	journal	December 2018
Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations Raissi, M.; Perdikaris, P.; Karniadakis, G. E. Journal of Computational Physics, Vol. 378 https://doi.org/10.1016/j.jcp.2018.10.045	journal	February 2019
Artificial neural networks for solving ordinary and partial differential equations Lagaris, I. E.; Likas, A.; Fotiadis, D. I. IEEE Transactions on Neural Networks, Vol. 9, Issue 5 https://doi.org/10.1109/72.712178	journal	January 1998
Image-to-Image Translation with Conditional Adversarial Networks Isola, Phillip; Zhu, Jun-Yan; Zhou, Tinghui 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) https://doi.org/10.1109/CVPR.2017.632	conference	July 2017
Generative Modeling Using the Sliced Wasserstein Distance Deshpande, Ishan; Zhang, Ziyu; Schwing, Alexander 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition https://doi.org/10.1109/CVPR.2018.00367	conference	June 2018
Deep Learning Face Attributes in the Wild Liu, Ziwei; Luo, Ping; Wang, Xiaogang 2015 IEEE International Conference on Computer Vision (ICCV) https://doi.org/10.1109/ICCV.2015.425	conference	December 2015
Unpaired Image-to-Image Translation Using Cycle-Consistent Adversarial Networks Zhu, Jun-Yan; Park, Taesung; Isola, Phillip 2017 IEEE International Conference on Computer Vision (ICCV) https://doi.org/10.1109/ICCV.2017.244	conference	October 2017
DualGAN: Unsupervised Dual Learning for Image-to-Image Translation Yi, Zili; Zhang, Hao; Tan, Ping 2017 IEEE International Conference on Computer Vision (ICCV) https://doi.org/10.1109/ICCV.2017.310	conference	October 2017
Inverting the Generator of a Generative Adversarial Network Creswell, Antonia; Bharath, Anil Anthony IEEE Transactions on Neural Networks and Learning Systems, Vol. 30, Issue 7 https://doi.org/10.1109/TNNLS.2018.2875194	journal	July 2019
Viscous Fingering: An Optimal Bound on the Growth Rate of the Mixing Zone Otto, Felix SIAM Journal on Applied Mathematics, Vol. 57, Issue 4 https://doi.org/10.1137/S003613999529438X	journal	August 1997

Similar Records

Physics-Informed Generative Adversarial Networks for Stochastic Differential Equations

Journal Article · Mon Feb 03 19:00:00 EST 2020 · SIAM Journal on Scientific Computing · OSTI ID:2281597

Emulation Modeling for Development of Cyber-Defense Capabilities for Satellite Systems

Conference · Tue Feb 22 23:00:00 EST 2022 · OSTI ID:1894014

Related Subjects

97 MATHEMATICS AND COMPUTING
generative adversarial networks (GANs)
generative models
image translation
normalizing flows
optimal transport
unpaired data

Potential Flow Generator With L2 Optimal Transport Regularity for Generative Models

Citation Formats

References (15)

Similar Records

Related Subjects

Potential Flow Generator With L₂ Optimal Transport Regularity for Generative Models