Please use this identifier to cite or link to this item: https://scidar.kg.ac.rs/handle/123456789/18700
Full metadata record
DC FieldValueLanguage
dc.contributor.authorKaplarević-Mališić, Ana-
dc.contributor.authorAndrijevic, Branka-
dc.contributor.authorBojovic, Filip-
dc.contributor.authorNikolić, Srđan-
dc.contributor.authorKrstic, Lazar-
dc.contributor.authorStojanović, Boban-
dc.contributor.authorIvanović, Miloš-
dc.date.accessioned2023-07-31T19:41:20Z-
dc.date.available2023-07-31T19:41:20Z-
dc.date.issued2023-
dc.identifier.urihttps://scidar.kg.ac.rs/handle/123456789/18700-
dc.description.abstractPhysics-Informed Neural Networks (PINNs) are artificial neural networks that encode Partial Differential Equations (PDEs) as an integral component of the ML model. PINNs are successfully used nowadays to solve PDEs, fractional equations, and integral–differential equations, including direct and inverse problems. Just as in the case of other kinds of artificial neural networks, the architecture, including the number and sizes of layers, activation functions, and other hyperparameters can significantly influence the network performance. Despite the serious work in this field, there are still no clear directions on how to choose an optimal network architecture in a consistent manner. In practice, expertise is required, with a significant number of manual trial and error cycles. In this paper, we propose PINN/GA (PINN/Genetic Algorithm), a fully automatic design of a PINN by an evolutionary strategy with specially tailored operators of selection, crossover, and mutation, adapted for deep neural network architecture and hyperparameter search. The PINN/GA strategy starts from the population of simple PINNs, adding new layers only if it brings clear accuracy benefits, keeping PINNs in the population as simple as possible. Since the examination of dozens of neural networks through the evolutionary process implies enormous computational costs, it employs a scalable computational design based on containers and Kubernetes batching orchestration. To demonstrate the potential of the proposed approach, we chose two non-trivial direct problems. The first is 1D Stefan transient model with time-dependent Dirichlet boundary conditions, describing the melting process, and the second is the Helmholtz wave equation over a 2D square domain. The authors found that PINNs accuracy gradually improves throughout the evolutionary process, exhibiting better performance and stability than parallel random search and Hyperopt Tree of Parzen Estimators, while keeping the network design reasonably simple.en_US
dc.language.isoen_USen_US
dc.publisherElsevieren_US
dc.rightsinfo:eu-repo/semantics/openAccess-
dc.sourceApplied Soft Computing-
dc.subjectPhysics Informed Neural Netowrksen_US
dc.subjectAutomatic designen_US
dc.subjectEvolutionary strategyen_US
dc.subjectGenetic Algorithmen_US
dc.titleIdentifying optimal architectures of physics-informed neural networks by evolutionary strategyen_US
dc.typearticleen_US
dc.description.versionAccepted for publishingen_US
dc.identifier.doi10.1016/j.asoc.2023.110646en_US
dc.type.versionReviewedVersionen_US
Appears in Collections:Faculty of Science, Kragujevac

Page views(s)

654

Downloads(s)

267

Files in This Item:
File Description SizeFormat 
STEFAN_PIN_GA-1.pdf1.38 MBAdobe PDFThumbnail
View/Open


Items in SCIDAR are protected by copyright, with all rights reserved, unless otherwise indicated.