{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,13]],"date-time":"2026-09-13T15:04:49Z","timestamp":1789311889570,"version":"build-2803163510"},"reference-count":52,"publisher":"Elsevier BV","license":[{"start":{"date-parts":[[2026,10,1]],"date-time":"2026-10-01T00:00:00Z","timestamp":1790812800000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/www.elsevier.com\/tdm\/userlicense\/1.0\/"},{"start":{"date-parts":[[2026,10,1]],"date-time":"2026-10-01T00:00:00Z","timestamp":1790812800000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/www.elsevier.com\/legal\/tdmrep-license"},{"start":{"date-parts":[[2026,10,1]],"date-time":"2026-10-01T00:00:00Z","timestamp":1790812800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-017"},{"start":{"date-parts":[[2026,10,1]],"date-time":"2026-10-01T00:00:00Z","timestamp":1790812800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"},{"start":{"date-parts":[[2026,10,1]],"date-time":"2026-10-01T00:00:00Z","timestamp":1790812800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-012"},{"start":{"date-parts":[[2026,10,1]],"date-time":"2026-10-01T00:00:00Z","timestamp":1790812800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2026,10,1]],"date-time":"2026-10-01T00:00:00Z","timestamp":1790812800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-004"}],"funder":[{"DOI":"10.13039\/501100010083","name":"Hunan Provincial Innovation Foundation for Postgraduate","doi-asserted-by":"publisher","id":[{"id":"10.13039\/501100010083","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["elsevier.com","sciencedirect.com"],"crossmark-restriction":true},"short-container-title":["Knowledge-Based Systems"],"published-print":{"date-parts":[[2026,10]]},"DOI":"10.1016\/j.knosys.2026.116586","type":"journal-article","created":{"date-parts":[[2026,7,9]],"date-time":"2026-07-09T20:16:36Z","timestamp":1783628196000},"page":"116586","update-policy":"https:\/\/doi.org\/10.1016\/elsevier_cm_policy","source":"Crossref","is-referenced-by-count":0,"special_numbering":"PA","title":["Dual-Gradient Co-optimization for Robust Physics-Informed Neural Networks"],"prefix":"10.1016","volume":"351","author":[{"given":"Zhuo","family":"Zhang","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Sen","family":"Zhang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Seongwon","family":"Kang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xi","family":"Yang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Dongseok","family":"Kim","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yifu","family":"Gao","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Canqun","family":"Yang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Chongam","family":"Kim","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"78","reference":[{"key":"10.1016\/j.knosys.2026.116586_b1","doi-asserted-by":"crossref","first-page":"686","DOI":"10.1016\/j.jcp.2018.10.045","article-title":"Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations","volume":"378","author":"Raissi","year":"2019","journal-title":"J. Comput. Phys."},{"issue":"6","key":"10.1016\/j.knosys.2026.116586_b2","doi-asserted-by":"crossref","first-page":"422","DOI":"10.1038\/s42254-021-00314-5","article-title":"Physics-informed machine learning","volume":"3","author":"Karniadakis","year":"2021","journal-title":"Nat. Rev. Phys."},{"key":"10.1016\/j.knosys.2026.116586_b3","unstructured":"N.H. Dashtbayaz, G. Farhani, B. Wang, C.X. Ling, Physics-Informed Neural Networks: Minimizing Residual Loss with Wide Networks and Effective Activations, in: IJCAI 2024."},{"issue":"2","key":"10.1016\/j.knosys.2026.116586_b4","doi-asserted-by":"crossref","first-page":"1111","DOI":"10.1007\/s00366-023-01861-4","article-title":"Developing an advanced neural network and physics solver coupled framework for accelerating flow field simulations","volume":"40","author":"Chen","year":"2024","journal-title":"Eng. Comput."},{"key":"10.1016\/j.knosys.2026.116586_b5","article-title":"A neural network approach for unstructured mesh quality evaluation","author":"Chen","year":"2024","journal-title":"Eng. Comput."},{"issue":"2","key":"10.1016\/j.knosys.2026.116586_b6","doi-asserted-by":"crossref","first-page":"207","DOI":"10.1631\/FITEE.2000435","article-title":"FlowDNN: a physics-informed deep neural network for fast and accurate flow prediction","volume":"23","author":"Chen","year":"2022","journal-title":"Front. Inf. Technol. Electron. Eng."},{"key":"10.1016\/j.knosys.2026.116586_b7","doi-asserted-by":"crossref","unstructured":"T. Matsubara, T. Yaguchi, Number Theoretic Accelerated Learning of Physics-Informed Neural Networks, in: Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 39, 2025, pp. 595\u2013603.","DOI":"10.1609\/aaai.v39i1.32040"},{"key":"10.1016\/j.knosys.2026.116586_b8","doi-asserted-by":"crossref","DOI":"10.1016\/j.knosys.2024.111831","article-title":"Unified finite-volume physics informed neural networks to solve the heterogeneous partial differential equations","volume":"295","author":"Mei","year":"2024","journal-title":"Knowl.-Based Syst."},{"key":"10.1016\/j.knosys.2026.116586_b9","article-title":"Bidirectional accelerated adaptive moment estimation for deep neural networks","author":"Lou","year":"2026","journal-title":"Knowl.-Based Syst."},{"key":"10.1016\/j.knosys.2026.116586_b10","article-title":"PI-MeshONet: A generalizable and self-supervised method for structured mesh generation","author":"Xiao","year":"2025","journal-title":"Knowl.-Based Syst."},{"key":"10.1016\/j.knosys.2026.116586_b11","doi-asserted-by":"crossref","DOI":"10.1016\/j.cma.2025.117956","article-title":"SK-PINN: Accelerated physics-informed deep learning by smoothing kernel gradients","volume":"440","author":"Pan","year":"2025","journal-title":"Comput. Methods Appl. Mech. Engrg."},{"key":"10.1016\/j.knosys.2026.116586_b12","doi-asserted-by":"crossref","DOI":"10.1016\/j.cma.2025.117751","article-title":"Frequency-adaptive multi-scale deep neural networks","volume":"437","author":"Huang","year":"2025","journal-title":"Comput. Methods Appl. Mech. Engrg."},{"key":"10.1016\/j.knosys.2026.116586_b13","series-title":"ICLR 2025, Singapore, April 24-28, 2025","article-title":"PINP: physics-informed neural predictor with latent estimation of fluid flows","author":"Chen","year":"2025"},{"key":"10.1016\/j.knosys.2026.116586_b14","unstructured":"Y. Li, S. Chen, B. Shan, S.-J. Huang, Causality-enhanced discreted physics-informed neural networks for predicting evolutionary equations, in: IJCAI 2024."},{"key":"10.1016\/j.knosys.2026.116586_b15","doi-asserted-by":"crossref","DOI":"10.1016\/j.engappai.2024.108496","article-title":"Optimized single-image super-resolution reconstruction: A multimodal approach based on reversible guidance and cyclical knowledge distillation","volume":"133","author":"Yan","year":"2024","journal-title":"Eng. Appl. Artif. Intell."},{"key":"10.1016\/j.knosys.2026.116586_b16","doi-asserted-by":"crossref","DOI":"10.1109\/TITS.2026.3680498","article-title":"Graph-MDETR: A graph-guided mamba-DETR network for UAV catenary support components detection in electrified railways","author":"Yang","year":"2026","journal-title":"IEEE Trans. Intell. Transp. Syst."},{"key":"10.1016\/j.knosys.2026.116586_b17","doi-asserted-by":"crossref","DOI":"10.1016\/j.engappai.2026.114691","article-title":"Spatial-channel collaborative multi-scale graph interaction deep transfer learning for unsupervised rotating machinery fault diagnosis","volume":"176","author":"Wang","year":"2026","journal-title":"Eng. Appl. Artif. Intell."},{"key":"10.1016\/j.knosys.2026.116586_b18","doi-asserted-by":"crossref","DOI":"10.1016\/j.jcp.2021.110768","article-title":"When and why PINNs fail to train: A neural tangent kernel perspective","volume":"449","author":"Wang","year":"2022","journal-title":"J. Comput. Phys."},{"issue":"3","key":"10.1016\/j.knosys.2026.116586_b19","article-title":"A pseudo-time stepping and parameterized physics-informed neural network framework for Navier\u2013Stokes equations","volume":"37","author":"Zhang","year":"2025","journal-title":"Phys. Fluids"},{"key":"10.1016\/j.knosys.2026.116586_b20","doi-asserted-by":"crossref","DOI":"10.1016\/j.jcp.2024.113494","article-title":"An analysis and solution of ill-conditioning in physics-informed neural networks","volume":"520","author":"Cao","year":"2025","journal-title":"J. Comput. Phys."},{"key":"10.1016\/j.knosys.2026.116586_b21","doi-asserted-by":"crossref","DOI":"10.1016\/j.jcp.2022.111121","article-title":"Meta-learning PINN loss functions","volume":"458","author":"Psaros","year":"2022","journal-title":"J. Comput. Phys."},{"key":"10.1016\/j.knosys.2026.116586_b22","doi-asserted-by":"crossref","DOI":"10.1016\/j.cma.2022.114823","article-title":"Gradient-enhanced physics-informed neural networks for forward and inverse PDE problems","volume":"393","author":"Yu","year":"2022","journal-title":"Comput. Methods Appl. Mech. Engrg."},{"key":"10.1016\/j.knosys.2026.116586_b23","doi-asserted-by":"crossref","DOI":"10.1016\/j.jcp.2023.112342","article-title":"Adaptive weighting of Bayesian physics informed neural networks for multitask and multiscale forward and inverse problems","volume":"491","author":"Perez","year":"2023","journal-title":"J. Comput. Phys."},{"key":"10.1016\/j.knosys.2026.116586_b24","doi-asserted-by":"crossref","first-page":"112","DOI":"10.1016\/j.neunet.2020.12.028","article-title":"A dual-dimer method for training physics-constrained neural networks with minimax architecture","volume":"136","author":"Liu","year":"2021","journal-title":"Neural Netw."},{"key":"10.1016\/j.knosys.2026.116586_b25","doi-asserted-by":"crossref","DOI":"10.1016\/j.jcp.2022.111722","article-title":"Self-adaptive physics-informed neural networks","volume":"474","author":"McClenny","year":"2023","journal-title":"J. Comput. Phys."},{"key":"10.1016\/j.knosys.2026.116586_b26","doi-asserted-by":"crossref","DOI":"10.1016\/j.jcp.2024.112781","article-title":"Loss-attentional physics-informed neural networks","volume":"501","author":"Song","year":"2024","journal-title":"J. Comput. Phys."},{"key":"10.1016\/j.knosys.2026.116586_b27","doi-asserted-by":"crossref","unstructured":"C. Zhou, J. Chen, Z. Yang, C.E. Png, Dual-Balancing for Physics-Informed Neural Networks, in: IJCAI 2025.","DOI":"10.24963\/ijcai.2025\/797"},{"key":"10.1016\/j.knosys.2026.116586_b28","series-title":"Respecting causality is all you need for training physics-informed neural networks","author":"Wang","year":"2022"},{"key":"10.1016\/j.knosys.2026.116586_b29","series-title":"On the Pareto front of physics-informed neural networks","author":"Rohrhofer","year":"2021"},{"key":"10.1016\/j.knosys.2026.116586_b30","doi-asserted-by":"crossref","DOI":"10.1016\/j.engappai.2025.110347","article-title":"Using physics-informed neural networks for solving Navier-Stokes equations in fluid dynamic complex scenarios","volume":"148","author":"Botarelli","year":"2025","journal-title":"Eng. Appl. Artif. Intell."},{"issue":"5","key":"10.1016\/j.knosys.2026.116586_b31","doi-asserted-by":"crossref","first-page":"A3055","DOI":"10.1137\/20M1318043","article-title":"Understanding and mitigating gradient flow pathologies in physics-informed neural networks","volume":"43","author":"Wang","year":"2021","journal-title":"SIAM J. Sci. Comput."},{"issue":"21","key":"10.1016\/j.knosys.2026.116586_b32","doi-asserted-by":"crossref","first-page":"8665","DOI":"10.3390\/s23218665","article-title":"Gradient statistics-based multi-objective optimization in physics-informed neural networks","volume":"23","author":"Vemuri","year":"2023","journal-title":"Sensors"},{"issue":"1","key":"10.1016\/j.knosys.2026.116586_b33","article-title":"Inverse Dirichlet weighting enables reliable training of physics informed neural networks","volume":"3","author":"Maddu","year":"2022","journal-title":"Mach. Learn.: Sci. Technol."},{"issue":"7","key":"10.1016\/j.knosys.2026.116586_b34","doi-asserted-by":"crossref","DOI":"10.1088\/2399-6528\/ace416","article-title":"Dynamic & norm-based weights to normalize imbalance in back-propagated gradients of physics-informed neural networks","volume":"7","author":"Deguchi","year":"2023","journal-title":"J. Phys. Commun."},{"key":"10.1016\/j.knosys.2026.116586_b35","series-title":"NeurIPS 2024, Vancouver, BC, Canada, December 10 - 15, 2024","article-title":"Dual cone gradient descent for training physics-informed neural networks","author":"Hwang","year":"2024"},{"key":"10.1016\/j.knosys.2026.116586_b36","series-title":"Advances in Neural Information Processing Systems","first-page":"110494","article-title":"Ropinn: Region optimized physics-informed neural networks","volume":"Vol. 37","author":"Wu","year":"2024"},{"key":"10.1016\/j.knosys.2026.116586_b37","article-title":"FIM-guided PINNs for real-time parameter identification in digital twins of nonlinear vibratory systems","author":"Banerjee","year":"2025","journal-title":"Knowl.-Based Syst."},{"key":"10.1016\/j.knosys.2026.116586_b38","doi-asserted-by":"crossref","DOI":"10.1016\/j.knosys.2025.113514","article-title":"Physics-informed neural networks prediction approach for knowledge dissemination model considering knowledge media","volume":"324","author":"Xia","year":"2025","journal-title":"Knowl.-Based Syst."},{"key":"10.1016\/j.knosys.2026.116586_b39","doi-asserted-by":"crossref","unstructured":"P. Ghanem, A. Demirkaya, T. Imbiriba, A. Ramezani, Z. Danziger, D. Erdogmus, Learning physics informed neural odes with partial measurements, in: Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 39, 2025, pp. 16799\u201316807.","DOI":"10.1609\/aaai.v39i16.33846"},{"key":"10.1016\/j.knosys.2026.116586_b40","series-title":"ICML 2024, Vienna, Austria, July 21-27, 2024","article-title":"Learning from integral losses in physics informed neural networks","author":"Saleh","year":"2024"},{"key":"10.1016\/j.knosys.2026.116586_b41","doi-asserted-by":"crossref","unstructured":"Y. Li, S.-C. Chen, S.-J. Huang, Implicit stochastic gradient descent for training physics-informed neural networks, in: Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 37, 2023, pp. 8692\u20138700.","DOI":"10.1609\/aaai.v37i7.26046"},{"key":"10.1016\/j.knosys.2026.116586_b42","doi-asserted-by":"crossref","unstructured":"Z. Yang, Z. Qiu, D. Fu, Dmis: Dynamic mesh-based importance sampling for training physics-informed neural networks, in: Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 37, 2023, pp. 5375\u20135383.","DOI":"10.1609\/aaai.v37i4.25669"},{"key":"10.1016\/j.knosys.2026.116586_b43","series-title":"Advances in Neural Information Processing Systems","article-title":"Kronecker-factored approximate curvature for physics-informed neural networks","author":"Dangel","year":"2024"},{"issue":"12","key":"10.1016\/j.knosys.2026.116586_b44","doi-asserted-by":"crossref","first-page":"15588","DOI":"10.1109\/TPAMI.2023.3307688","article-title":"Self-scalable tanh (stan): Multi-scale solutions for physics-informed neural networks","volume":"45","author":"Gnanasambandam","year":"2023","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"10.1016\/j.knosys.2026.116586_b45","series-title":"ICML 2024, Vienna, Austria, July 21-27, 2024","article-title":"Parameterized physics-informed neural networks for parameterized PDEs","author":"Cho","year":"2024"},{"key":"10.1016\/j.knosys.2026.116586_b46","doi-asserted-by":"crossref","unstructured":"T. Kapoor, A. Chandra, D.M. Tartakovsky, H. Wang, A. Nunez, R. Dollevoet, Neural oscillators for generalization of physics-informed machine learning, in: Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 38, 2024, pp. 13059\u201313067.","DOI":"10.1609\/aaai.v38i12.29204"},{"key":"10.1016\/j.knosys.2026.116586_b47","doi-asserted-by":"crossref","unstructured":"I. Ashraf, J. Strotherm, L. Hermes, B. Hammer, Physics-informed graph neural networks for water distribution systems, in: Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 38, 2024, pp. 21905\u201321913.","DOI":"10.1609\/aaai.v38i20.30192"},{"key":"10.1016\/j.knosys.2026.116586_b48","doi-asserted-by":"crossref","first-page":"11","DOI":"10.1016\/j.neucom.2022.05.015","article-title":"Self-adaptive loss balanced physics-informed neural networks","volume":"496","author":"Xiang","year":"2022","journal-title":"Neurocomputing"},{"key":"10.1016\/j.knosys.2026.116586_b49","doi-asserted-by":"crossref","first-page":"86252","DOI":"10.1109\/ACCESS.2023.3302892","article-title":"Data vs. physics: The apparent pareto front of physics-informed neural networks","volume":"11","author":"Rohrhofer","year":"2023","journal-title":"IEEE Access"},{"key":"10.1016\/j.knosys.2026.116586_b50","series-title":"Advances in Neural Information Processing Systems","first-page":"26548","article-title":"Characterizing possible failure modes in physics-informed neural networks","volume":"Vol. 34","author":"Krishnapriyan","year":"2021"},{"key":"10.1016\/j.knosys.2026.116586_b51","series-title":"Advances in Neural Information Processing Systems","first-page":"7537","article-title":"Fourier features let networks learn high frequency functions in low dimensional domains","volume":"Vol. 33","author":"Tancik","year":"2020"},{"issue":"2239","key":"10.1016\/j.knosys.2026.116586_b52","doi-asserted-by":"crossref","DOI":"10.1098\/rspa.2020.0334","article-title":"Locally adaptive activation functions with slope recovery for deep and physics-informed neural networks","volume":"476","author":"Jagtap","year":"2020","journal-title":"Proc. R. Soc. A"}],"container-title":["Knowledge-Based Systems"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/api.elsevier.com\/content\/article\/PII:S0950705126013122?httpAccept=text\/xml","content-type":"text\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/api.elsevier.com\/content\/article\/PII:S0950705126013122?httpAccept=text\/plain","content-type":"text\/plain","content-version":"vor","intended-application":"text-mining"}],"deposited":{"date-parts":[[2026,9,13]],"date-time":"2026-09-13T14:50:02Z","timestamp":1789311002000},"score":1,"resource":{"primary":{"URL":"https:\/\/linkinghub.elsevier.com\/retrieve\/pii\/S0950705126013122"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,10]]},"references-count":52,"alternative-id":["S0950705126013122"],"URL":"https:\/\/doi.org\/10.1016\/j.knosys.2026.116586","relation":{},"ISSN":["0950-7051"],"issn-type":[{"value":"0950-7051","type":"print"}],"subject":[],"published":{"date-parts":[[2026,10]]},"assertion":[{"value":"Elsevier","name":"publisher","label":"This article is maintained by"},{"value":"Dual-Gradient Co-optimization for Robust Physics-Informed Neural Networks","name":"articletitle","label":"Article Title"},{"value":"Knowledge-Based Systems","name":"journaltitle","label":"Journal Title"},{"value":"https:\/\/doi.org\/10.1016\/j.knosys.2026.116586","name":"articlelink","label":"CrossRef DOI link to publisher maintained version"},{"value":"article","name":"content_type","label":"Content Type"},{"value":"\u00a9 2026 Elsevier B.V. All rights are reserved, including those for text and data mining, AI training, and similar technologies.","name":"copyright","label":"Copyright"}],"article-number":"116586"}}