{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,4,22]],"date-time":"2026-04-22T20:15:33Z","timestamp":1776888933281,"version":"3.51.2"},"reference-count":21,"publisher":"IOP Publishing","issue":"2","license":[{"start":{"date-parts":[[2025,4,16]],"date-time":"2025-04-16T00:00:00Z","timestamp":1744761600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"},{"start":{"date-parts":[[2025,4,16]],"date-time":"2025-04-16T00:00:00Z","timestamp":1744761600000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/iopscience.iop.org\/info\/page\/text-and-data-mining"}],"funder":[{"DOI":"10.13039\/100006209","name":"Nuclear Physics","doi-asserted-by":"crossref","award":["DE-AC05-06OR23177"],"award-info":[{"award-number":["DE-AC05-06OR23177"]}],"id":[{"id":"10.13039\/100006209","id-type":"DOI","asserted-by":"crossref"}]},{"DOI":"10.13039\/100006192","name":"Advanced Scientific Computing Research","doi-asserted-by":"crossref","award":["DE-AC02-06CH11357"],"award-info":[{"award-number":["DE-AC02-06CH11357"]}],"id":[{"id":"10.13039\/100006192","id-type":"DOI","asserted-by":"crossref"}]}],"content-domain":{"domain":["iopscience.iop.org"],"crossmark-restriction":false},"short-container-title":["Mach. Learn.: Sci. Technol."],"published-print":{"date-parts":[[2025,6,30]]},"abstract":"<jats:title>Abstract<\/jats:title>\n               <jats:p>Solving large-scale inverse problems using deep-learning algorithms have become an essential part of modern research and industrial applications. The complexity of the underlying inverse problem may require the utilization of high performance computing systems which poses a challenge on the algorithmic design of the inverse problem solver. Most deep learning algorithms require, due to their design, custom parallelization techniques in order to be resource efficient while showing a reasonable convergence. In this paper we introduce a <jats:underline>S<\/jats:underline>calable <jats:underline>A<\/jats:underline>synchronous <jats:underline>G<\/jats:underline>enerative <jats:underline>I<\/jats:underline>nverse <jats:underline>P<\/jats:underline>roblem <jats:underline>S<\/jats:underline>olver (SAGIPS) on high-performance computing systems. We present a workflow that utilizes an asynchronous ring-allreduce algorithm to transfer the gradients of the generator network across multiple GPUs. Experiments with a scientific proxy application demonstrate that SAGIPS shows near linear weak scaling, together with a convergence quality that is comparable to traditional methods. The approach presented here allows leveraging Generative Adverserial Network across multiple GPUs, promising advancements in solving complex inverse problems at scale.<\/jats:p>","DOI":"10.1088\/2632-2153\/adc8fb","type":"journal-article","created":{"date-parts":[[2025,4,4]],"date-time":"2025-04-04T08:57:54Z","timestamp":1743757074000},"page":"025017","update-policy":"https:\/\/doi.org\/10.1088\/crossmark-policy","source":"Crossref","is-referenced-by-count":2,"title":["SAGIPS: a physics-inspired scalable asynchronous generative inverse-problem solver"],"prefix":"10.1088","volume":"6","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-0356-0754","authenticated-orcid":true,"given":"Daniel","family":"Lersch","sequence":"first","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3475-2871","authenticated-orcid":true,"given":"Malachi","family":"Schram","sequence":"additional","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6135-7749","authenticated-orcid":true,"given":"Zhenyu","family":"Dai","sequence":"additional","affiliation":[]},{"given":"Kishansingh","family":"Rajput","sequence":"additional","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0002-1535-6208","authenticated-orcid":true,"given":"Nobuo","family":"Sato","sequence":"additional","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0001-8150-5171","authenticated-orcid":true,"given":"Xingfu","family":"Wu","sequence":"additional","affiliation":[]},{"given":"J Taylor","family":"Childers","sequence":"additional","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5264-6298","authenticated-orcid":true,"given":"Steven","family":"Goldenberg","sequence":"additional","affiliation":[]}],"member":"266","published-online":{"date-parts":[[2025,4,16]]},"reference":[{"key":"mlstadc8fbbib1","first-page":"pp 2234","article-title":"Improved techniques for training GANs","author":"Salimans","year":"2016"},{"key":"mlstadc8fbbib2","doi-asserted-by":"publisher","first-page":"45559","DOI":"10.1109\/ACCESS.2023.3274201","article-title":"Computationally efficient neural rendering for generator adversarial networks using a multi-GPU cluster in a cloud environment","volume":"11","author":"Ravikumar","year":"2023","journal-title":"IEEE Access"},{"key":"mlstadc8fbbib3","first-page":"2459","article-title":"MGAN: training generative adversarial nets with multiple generators","volume":"4","author":"Hoang","year":"2018","journal-title":"ICLR"},{"key":"mlstadc8fbbib4","doi-asserted-by":"publisher","first-page":"193","DOI":"10.1145\/3423211.3425688","article-title":"FeGAN: scaling distributed GANs","author":"Guerraoui","year":"2020"},{"key":"mlstadc8fbbib5","first-page":"pp 866","article-title":"MD-GAN: multi-discriminator generative adversarial networks for distributed datasets","author":"Hardy","year":"2019"},{"key":"mlstadc8fbbib6","first-page":"pp 2672","article-title":"Generative adversarial nets","volume":"vol 2","author":"Goodfellow","year":"2014"},{"key":"mlstadc8fbbib7","first-page":"2672","article-title":"Conditional generative adversarial nets","author":"Mirza","year":"2014","journal-title":"Computer Science"},{"key":"mlstadc8fbbib8","article-title":"PyTorch: an imperative style, high-performance deep learning library","author":"Paszke","year":"2019"},{"key":"mlstadc8fbbib9","first-page":"pp 6405","article-title":"Simple and scalable predictive uncertainty estimation using deep ensembles","author":"Lakshminarayanan","year":"2017"},{"key":"mlstadc8fbbib10","first-page":"pp 5430","article-title":"AdaGAN: boosting generative models","author":"Tolstikhin","year":"2017"},{"key":"mlstadc8fbbib11","doi-asserted-by":"publisher","first-page":"4090","DOI":"10.1609\/aaai.v35i5.16530","article-title":"GAN ensemble for anomaly detection","volume":"vol 5","author":"Han","year":"2021"},{"key":"mlstadc8fbbib12","article-title":"Highly scalable deep learning training system with mixed-precision: Training imagenet in four minutes","author":"Jia","year":"2018"},{"key":"mlstadc8fbbib13","article-title":"Massively distributed SGD: ImageNet\/ResNet-50 training in a flash","author":"Mikami","year":"2019"},{"key":"mlstadc8fbbib14","doi-asserted-by":"publisher","first-page":"581","DOI":"10.1016\/j.parco.2009.09.001","article-title":"Two-tree algorithms for full bandwidth broadcast, reduction and scan","volume":"35","author":"Sanders","year":"2009"},{"key":"mlstadc8fbbib15","first-page":"pp 173","article-title":"Deep speech 2: end-to-end speech recognition in English and Mandarin","author":"Amodei","year":"2016"},{"key":"mlstadc8fbbib16","doi-asserted-by":"publisher","first-page":"47","DOI":"10.1109\/MCSE.2021.3083216","article-title":"mpi4py: status update after 12 years of development","volume":"23","author":"Dalcin","year":"2021","journal-title":"Comput. Sci. Eng."},{"key":"mlstadc8fbbib17","doi-asserted-by":"publisher","first-page":"3005","DOI":"10.14778\/3415478.3415530","article-title":"PyTorch distributed: experiences on accelerating data parallel training","volume":"13","author":"Li","year":"2020","journal-title":"Proc. VLDB Endow."},{"key":"mlstadc8fbbib18","first-page":"pp 265","article-title":"TensorFlow: a system for large-scale machine learning","author":"Abadi","year":"2016"},{"key":"mlstadc8fbbib19","article-title":"Rectifier nonlinearities improve neural network acoustic models","author":"Maas","year":"2013"},{"key":"mlstadc8fbbib20","article-title":"Performance, power and scalability analysis of the Horovod implementation of the CANDLE NT3 benchmark on the Cray XC40 Theta SC18","author":"Wu","year":"2018"},{"key":"mlstadc8fbbib21","doi-asserted-by":"publisher","DOI":"10.1145\/3337821.3337905","article-title":"Performance, energy and scalability analysis and improvement of parallel cancer deep learning candle benchmarks","author":"Wu","year":"2019"}],"container-title":["Machine Learning: Science and Technology"],"original-title":[],"link":[{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/adc8fb","content-type":"text\/html","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/adc8fb\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/adc8fb","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/adc8fb\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/adc8fb\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/adc8fb\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/adc8fb\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"similarity-checking"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/adc8fb\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,4,16]],"date-time":"2025-04-16T10:26:39Z","timestamp":1744799199000},"score":1,"resource":{"primary":{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/adc8fb"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,4,16]]},"references-count":21,"journal-issue":{"issue":"2","published-online":{"date-parts":[[2025,4,16]]},"published-print":{"date-parts":[[2025,6,30]]}},"URL":"https:\/\/doi.org\/10.1088\/2632-2153\/adc8fb","relation":{},"ISSN":["2632-2153"],"issn-type":[{"value":"2632-2153","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,4,16]]},"assertion":[{"value":"SAGIPS: a physics-inspired scalable asynchronous generative inverse-problem solver","name":"article_title","label":"Article Title"},{"value":"Machine Learning: Science and Technology","name":"journal_title","label":"Journal Title"},{"value":"paper","name":"article_type","label":"Article Type"},{"value":"\u00a9 2025 The Author(s). Published by IOP Publishing Ltd. Contribution of National Oceanic and Atmospheric Administration is not subject to copyright in the USA","name":"copyright_information","label":"Copyright Information"},{"value":"2024-11-01","name":"date_received","label":"Date Received","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2025-04-03","name":"date_accepted","label":"Date Accepted","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2025-04-16","name":"date_epub","label":"Online publication date","group":{"name":"publication_dates","label":"Publication dates"}}]}}