BC-PINN: an adaptive physics informed neural network based on biased multiobjective coevolutionary algorithm
Abstract Physics informed neural network (PINN) has become a promising method for solving partial differential equations (PDEs). The loss function of PINN is a weighted sum of multiple items. This makes it easy to fall into local optima, especially the gradient pathologies when solving high frequenc...
Ausführliche Beschreibung
Autor*in: |
Zhu, Zhicheng [verfasserIn] |
---|
Format: |
Artikel |
---|---|
Sprache: |
Englisch |
Erschienen: |
2023 |
---|
Schlagwörter: |
---|
Anmerkung: |
© The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2023. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. |
---|
Übergeordnetes Werk: |
Enthalten in: Neural computing & applications - Springer London, 1993, 35(2023), 28 vom: 02. Aug., Seite 21093-21113 |
---|---|
Übergeordnetes Werk: |
volume:35 ; year:2023 ; number:28 ; day:02 ; month:08 ; pages:21093-21113 |
Links: |
---|
DOI / URN: |
10.1007/s00521-023-08876-4 |
---|
Katalog-ID: |
OLC2145282378 |
---|
LEADER | 01000naa a22002652 4500 | ||
---|---|---|---|
001 | OLC2145282378 | ||
003 | DE-627 | ||
005 | 20240118103717.0 | ||
007 | tu | ||
008 | 240118s2023 xx ||||| 00| ||eng c | ||
024 | 7 | |a 10.1007/s00521-023-08876-4 |2 doi | |
035 | |a (DE-627)OLC2145282378 | ||
035 | |a (DE-He213)s00521-023-08876-4-p | ||
040 | |a DE-627 |b ger |c DE-627 |e rakwb | ||
041 | |a eng | ||
082 | 0 | 4 | |a 004 |q VZ |
100 | 1 | |a Zhu, Zhicheng |e verfasserin |4 aut | |
245 | 1 | 0 | |a BC-PINN: an adaptive physics informed neural network based on biased multiobjective coevolutionary algorithm |
264 | 1 | |c 2023 | |
336 | |a Text |b txt |2 rdacontent | ||
337 | |a ohne Hilfsmittel zu benutzen |b n |2 rdamedia | ||
338 | |a Band |b nc |2 rdacarrier | ||
500 | |a © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2023. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. | ||
520 | |a Abstract Physics informed neural network (PINN) has become a promising method for solving partial differential equations (PDEs). The loss function of PINN is a weighted sum of multiple items. This makes it easy to fall into local optima, especially the gradient pathologies when solving high frequency problems. The value of penalty coefficients has a crucial impact on the prediction results. Therefore, a new PINN with adaptive penalty coefficients iteratively optimized by biased multiobjective coevolutionary algorithm (BC-PINN) is presented. In BC-PINN, a two-stage optimization mechanism is used to search for parameters of neural network and penalty coefficients respectively. This method involves constructing the fitness function of penalty coefficients based on the biased dominance ranking by data item and regularization item. Compared with the previous works of others, the accuracy of fitting the initial conditions and boundary conditions is considered to be given priority, which is more conducive to PINN converging to the particular solution of PDE. In addition, the set of penalty coefficients is divided into multiple populations to improve the optimization efficiency through coevolutionary algorithm. The empirical results show that: (1) Our method can improve the gradient pathologies and effectively capture the high-frequency features. (2) Compared to the original PINN, it reduces the MSE by 1–6 orders of magnitude in our benchmark functions. | ||
650 | 4 | |a Physics informed neural network | |
650 | 4 | |a Gradient pathologies | |
650 | 4 | |a Coevolutionary algorithm | |
650 | 4 | |a Biased multiobjective optimization | |
700 | 1 | |a Hao, Jia |4 aut | |
700 | 1 | |a Huang, Jin |4 aut | |
700 | 1 | |a Huang, Biao |4 aut | |
773 | 0 | 8 | |i Enthalten in |t Neural computing & applications |d Springer London, 1993 |g 35(2023), 28 vom: 02. Aug., Seite 21093-21113 |w (DE-627)165669608 |w (DE-600)1136944-9 |w (DE-576)032873050 |x 0941-0643 |7 nnns |
773 | 1 | 8 | |g volume:35 |g year:2023 |g number:28 |g day:02 |g month:08 |g pages:21093-21113 |
856 | 4 | 1 | |u https://doi.org/10.1007/s00521-023-08876-4 |z lizenzpflichtig |3 Volltext |
912 | |a GBV_USEFLAG_A | ||
912 | |a SYSFLAG_A | ||
912 | |a GBV_OLC | ||
912 | |a SSG-OLC-MAT | ||
912 | |a GBV_ILN_2018 | ||
912 | |a GBV_ILN_4277 | ||
951 | |a AR | ||
952 | |d 35 |j 2023 |e 28 |b 02 |c 08 |h 21093-21113 |
author_variant |
z z zz j h jh j h jh b h bh |
---|---|
matchkey_str |
article:09410643:2023----::cinndpiehscifrenuantokaeobaemlibet |
hierarchy_sort_str |
2023 |
publishDate |
2023 |
allfields |
10.1007/s00521-023-08876-4 doi (DE-627)OLC2145282378 (DE-He213)s00521-023-08876-4-p DE-627 ger DE-627 rakwb eng 004 VZ Zhu, Zhicheng verfasserin aut BC-PINN: an adaptive physics informed neural network based on biased multiobjective coevolutionary algorithm 2023 Text txt rdacontent ohne Hilfsmittel zu benutzen n rdamedia Band nc rdacarrier © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2023. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. Abstract Physics informed neural network (PINN) has become a promising method for solving partial differential equations (PDEs). The loss function of PINN is a weighted sum of multiple items. This makes it easy to fall into local optima, especially the gradient pathologies when solving high frequency problems. The value of penalty coefficients has a crucial impact on the prediction results. Therefore, a new PINN with adaptive penalty coefficients iteratively optimized by biased multiobjective coevolutionary algorithm (BC-PINN) is presented. In BC-PINN, a two-stage optimization mechanism is used to search for parameters of neural network and penalty coefficients respectively. This method involves constructing the fitness function of penalty coefficients based on the biased dominance ranking by data item and regularization item. Compared with the previous works of others, the accuracy of fitting the initial conditions and boundary conditions is considered to be given priority, which is more conducive to PINN converging to the particular solution of PDE. In addition, the set of penalty coefficients is divided into multiple populations to improve the optimization efficiency through coevolutionary algorithm. The empirical results show that: (1) Our method can improve the gradient pathologies and effectively capture the high-frequency features. (2) Compared to the original PINN, it reduces the MSE by 1–6 orders of magnitude in our benchmark functions. Physics informed neural network Gradient pathologies Coevolutionary algorithm Biased multiobjective optimization Hao, Jia aut Huang, Jin aut Huang, Biao aut Enthalten in Neural computing & applications Springer London, 1993 35(2023), 28 vom: 02. Aug., Seite 21093-21113 (DE-627)165669608 (DE-600)1136944-9 (DE-576)032873050 0941-0643 nnns volume:35 year:2023 number:28 day:02 month:08 pages:21093-21113 https://doi.org/10.1007/s00521-023-08876-4 lizenzpflichtig Volltext GBV_USEFLAG_A SYSFLAG_A GBV_OLC SSG-OLC-MAT GBV_ILN_2018 GBV_ILN_4277 AR 35 2023 28 02 08 21093-21113 |
spelling |
10.1007/s00521-023-08876-4 doi (DE-627)OLC2145282378 (DE-He213)s00521-023-08876-4-p DE-627 ger DE-627 rakwb eng 004 VZ Zhu, Zhicheng verfasserin aut BC-PINN: an adaptive physics informed neural network based on biased multiobjective coevolutionary algorithm 2023 Text txt rdacontent ohne Hilfsmittel zu benutzen n rdamedia Band nc rdacarrier © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2023. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. Abstract Physics informed neural network (PINN) has become a promising method for solving partial differential equations (PDEs). The loss function of PINN is a weighted sum of multiple items. This makes it easy to fall into local optima, especially the gradient pathologies when solving high frequency problems. The value of penalty coefficients has a crucial impact on the prediction results. Therefore, a new PINN with adaptive penalty coefficients iteratively optimized by biased multiobjective coevolutionary algorithm (BC-PINN) is presented. In BC-PINN, a two-stage optimization mechanism is used to search for parameters of neural network and penalty coefficients respectively. This method involves constructing the fitness function of penalty coefficients based on the biased dominance ranking by data item and regularization item. Compared with the previous works of others, the accuracy of fitting the initial conditions and boundary conditions is considered to be given priority, which is more conducive to PINN converging to the particular solution of PDE. In addition, the set of penalty coefficients is divided into multiple populations to improve the optimization efficiency through coevolutionary algorithm. The empirical results show that: (1) Our method can improve the gradient pathologies and effectively capture the high-frequency features. (2) Compared to the original PINN, it reduces the MSE by 1–6 orders of magnitude in our benchmark functions. Physics informed neural network Gradient pathologies Coevolutionary algorithm Biased multiobjective optimization Hao, Jia aut Huang, Jin aut Huang, Biao aut Enthalten in Neural computing & applications Springer London, 1993 35(2023), 28 vom: 02. Aug., Seite 21093-21113 (DE-627)165669608 (DE-600)1136944-9 (DE-576)032873050 0941-0643 nnns volume:35 year:2023 number:28 day:02 month:08 pages:21093-21113 https://doi.org/10.1007/s00521-023-08876-4 lizenzpflichtig Volltext GBV_USEFLAG_A SYSFLAG_A GBV_OLC SSG-OLC-MAT GBV_ILN_2018 GBV_ILN_4277 AR 35 2023 28 02 08 21093-21113 |
allfields_unstemmed |
10.1007/s00521-023-08876-4 doi (DE-627)OLC2145282378 (DE-He213)s00521-023-08876-4-p DE-627 ger DE-627 rakwb eng 004 VZ Zhu, Zhicheng verfasserin aut BC-PINN: an adaptive physics informed neural network based on biased multiobjective coevolutionary algorithm 2023 Text txt rdacontent ohne Hilfsmittel zu benutzen n rdamedia Band nc rdacarrier © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2023. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. Abstract Physics informed neural network (PINN) has become a promising method for solving partial differential equations (PDEs). The loss function of PINN is a weighted sum of multiple items. This makes it easy to fall into local optima, especially the gradient pathologies when solving high frequency problems. The value of penalty coefficients has a crucial impact on the prediction results. Therefore, a new PINN with adaptive penalty coefficients iteratively optimized by biased multiobjective coevolutionary algorithm (BC-PINN) is presented. In BC-PINN, a two-stage optimization mechanism is used to search for parameters of neural network and penalty coefficients respectively. This method involves constructing the fitness function of penalty coefficients based on the biased dominance ranking by data item and regularization item. Compared with the previous works of others, the accuracy of fitting the initial conditions and boundary conditions is considered to be given priority, which is more conducive to PINN converging to the particular solution of PDE. In addition, the set of penalty coefficients is divided into multiple populations to improve the optimization efficiency through coevolutionary algorithm. The empirical results show that: (1) Our method can improve the gradient pathologies and effectively capture the high-frequency features. (2) Compared to the original PINN, it reduces the MSE by 1–6 orders of magnitude in our benchmark functions. Physics informed neural network Gradient pathologies Coevolutionary algorithm Biased multiobjective optimization Hao, Jia aut Huang, Jin aut Huang, Biao aut Enthalten in Neural computing & applications Springer London, 1993 35(2023), 28 vom: 02. Aug., Seite 21093-21113 (DE-627)165669608 (DE-600)1136944-9 (DE-576)032873050 0941-0643 nnns volume:35 year:2023 number:28 day:02 month:08 pages:21093-21113 https://doi.org/10.1007/s00521-023-08876-4 lizenzpflichtig Volltext GBV_USEFLAG_A SYSFLAG_A GBV_OLC SSG-OLC-MAT GBV_ILN_2018 GBV_ILN_4277 AR 35 2023 28 02 08 21093-21113 |
allfieldsGer |
10.1007/s00521-023-08876-4 doi (DE-627)OLC2145282378 (DE-He213)s00521-023-08876-4-p DE-627 ger DE-627 rakwb eng 004 VZ Zhu, Zhicheng verfasserin aut BC-PINN: an adaptive physics informed neural network based on biased multiobjective coevolutionary algorithm 2023 Text txt rdacontent ohne Hilfsmittel zu benutzen n rdamedia Band nc rdacarrier © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2023. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. Abstract Physics informed neural network (PINN) has become a promising method for solving partial differential equations (PDEs). The loss function of PINN is a weighted sum of multiple items. This makes it easy to fall into local optima, especially the gradient pathologies when solving high frequency problems. The value of penalty coefficients has a crucial impact on the prediction results. Therefore, a new PINN with adaptive penalty coefficients iteratively optimized by biased multiobjective coevolutionary algorithm (BC-PINN) is presented. In BC-PINN, a two-stage optimization mechanism is used to search for parameters of neural network and penalty coefficients respectively. This method involves constructing the fitness function of penalty coefficients based on the biased dominance ranking by data item and regularization item. Compared with the previous works of others, the accuracy of fitting the initial conditions and boundary conditions is considered to be given priority, which is more conducive to PINN converging to the particular solution of PDE. In addition, the set of penalty coefficients is divided into multiple populations to improve the optimization efficiency through coevolutionary algorithm. The empirical results show that: (1) Our method can improve the gradient pathologies and effectively capture the high-frequency features. (2) Compared to the original PINN, it reduces the MSE by 1–6 orders of magnitude in our benchmark functions. Physics informed neural network Gradient pathologies Coevolutionary algorithm Biased multiobjective optimization Hao, Jia aut Huang, Jin aut Huang, Biao aut Enthalten in Neural computing & applications Springer London, 1993 35(2023), 28 vom: 02. Aug., Seite 21093-21113 (DE-627)165669608 (DE-600)1136944-9 (DE-576)032873050 0941-0643 nnns volume:35 year:2023 number:28 day:02 month:08 pages:21093-21113 https://doi.org/10.1007/s00521-023-08876-4 lizenzpflichtig Volltext GBV_USEFLAG_A SYSFLAG_A GBV_OLC SSG-OLC-MAT GBV_ILN_2018 GBV_ILN_4277 AR 35 2023 28 02 08 21093-21113 |
allfieldsSound |
10.1007/s00521-023-08876-4 doi (DE-627)OLC2145282378 (DE-He213)s00521-023-08876-4-p DE-627 ger DE-627 rakwb eng 004 VZ Zhu, Zhicheng verfasserin aut BC-PINN: an adaptive physics informed neural network based on biased multiobjective coevolutionary algorithm 2023 Text txt rdacontent ohne Hilfsmittel zu benutzen n rdamedia Band nc rdacarrier © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2023. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. Abstract Physics informed neural network (PINN) has become a promising method for solving partial differential equations (PDEs). The loss function of PINN is a weighted sum of multiple items. This makes it easy to fall into local optima, especially the gradient pathologies when solving high frequency problems. The value of penalty coefficients has a crucial impact on the prediction results. Therefore, a new PINN with adaptive penalty coefficients iteratively optimized by biased multiobjective coevolutionary algorithm (BC-PINN) is presented. In BC-PINN, a two-stage optimization mechanism is used to search for parameters of neural network and penalty coefficients respectively. This method involves constructing the fitness function of penalty coefficients based on the biased dominance ranking by data item and regularization item. Compared with the previous works of others, the accuracy of fitting the initial conditions and boundary conditions is considered to be given priority, which is more conducive to PINN converging to the particular solution of PDE. In addition, the set of penalty coefficients is divided into multiple populations to improve the optimization efficiency through coevolutionary algorithm. The empirical results show that: (1) Our method can improve the gradient pathologies and effectively capture the high-frequency features. (2) Compared to the original PINN, it reduces the MSE by 1–6 orders of magnitude in our benchmark functions. Physics informed neural network Gradient pathologies Coevolutionary algorithm Biased multiobjective optimization Hao, Jia aut Huang, Jin aut Huang, Biao aut Enthalten in Neural computing & applications Springer London, 1993 35(2023), 28 vom: 02. Aug., Seite 21093-21113 (DE-627)165669608 (DE-600)1136944-9 (DE-576)032873050 0941-0643 nnns volume:35 year:2023 number:28 day:02 month:08 pages:21093-21113 https://doi.org/10.1007/s00521-023-08876-4 lizenzpflichtig Volltext GBV_USEFLAG_A SYSFLAG_A GBV_OLC SSG-OLC-MAT GBV_ILN_2018 GBV_ILN_4277 AR 35 2023 28 02 08 21093-21113 |
language |
English |
source |
Enthalten in Neural computing & applications 35(2023), 28 vom: 02. Aug., Seite 21093-21113 volume:35 year:2023 number:28 day:02 month:08 pages:21093-21113 |
sourceStr |
Enthalten in Neural computing & applications 35(2023), 28 vom: 02. Aug., Seite 21093-21113 volume:35 year:2023 number:28 day:02 month:08 pages:21093-21113 |
format_phy_str_mv |
Article |
institution |
findex.gbv.de |
topic_facet |
Physics informed neural network Gradient pathologies Coevolutionary algorithm Biased multiobjective optimization |
dewey-raw |
004 |
isfreeaccess_bool |
false |
container_title |
Neural computing & applications |
authorswithroles_txt_mv |
Zhu, Zhicheng @@aut@@ Hao, Jia @@aut@@ Huang, Jin @@aut@@ Huang, Biao @@aut@@ |
publishDateDaySort_date |
2023-08-02T00:00:00Z |
hierarchy_top_id |
165669608 |
dewey-sort |
14 |
id |
OLC2145282378 |
language_de |
englisch |
fullrecord |
<?xml version="1.0" encoding="UTF-8"?><collection xmlns="http://www.loc.gov/MARC21/slim"><record><leader>01000naa a22002652 4500</leader><controlfield tag="001">OLC2145282378</controlfield><controlfield tag="003">DE-627</controlfield><controlfield tag="005">20240118103717.0</controlfield><controlfield tag="007">tu</controlfield><controlfield tag="008">240118s2023 xx ||||| 00| ||eng c</controlfield><datafield tag="024" ind1="7" ind2=" "><subfield code="a">10.1007/s00521-023-08876-4</subfield><subfield code="2">doi</subfield></datafield><datafield tag="035" ind1=" " ind2=" "><subfield code="a">(DE-627)OLC2145282378</subfield></datafield><datafield tag="035" ind1=" " ind2=" "><subfield code="a">(DE-He213)s00521-023-08876-4-p</subfield></datafield><datafield tag="040" ind1=" " ind2=" "><subfield code="a">DE-627</subfield><subfield code="b">ger</subfield><subfield code="c">DE-627</subfield><subfield code="e">rakwb</subfield></datafield><datafield tag="041" ind1=" " ind2=" "><subfield code="a">eng</subfield></datafield><datafield tag="082" ind1="0" ind2="4"><subfield code="a">004</subfield><subfield code="q">VZ</subfield></datafield><datafield tag="100" ind1="1" ind2=" "><subfield code="a">Zhu, Zhicheng</subfield><subfield code="e">verfasserin</subfield><subfield code="4">aut</subfield></datafield><datafield tag="245" ind1="1" ind2="0"><subfield code="a">BC-PINN: an adaptive physics informed neural network based on biased multiobjective coevolutionary algorithm</subfield></datafield><datafield tag="264" ind1=" " ind2="1"><subfield code="c">2023</subfield></datafield><datafield tag="336" ind1=" " ind2=" "><subfield code="a">Text</subfield><subfield code="b">txt</subfield><subfield code="2">rdacontent</subfield></datafield><datafield tag="337" ind1=" " ind2=" "><subfield code="a">ohne Hilfsmittel zu benutzen</subfield><subfield code="b">n</subfield><subfield code="2">rdamedia</subfield></datafield><datafield tag="338" ind1=" " ind2=" "><subfield code="a">Band</subfield><subfield code="b">nc</subfield><subfield code="2">rdacarrier</subfield></datafield><datafield tag="500" ind1=" " ind2=" "><subfield code="a">© The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2023. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.</subfield></datafield><datafield tag="520" ind1=" " ind2=" "><subfield code="a">Abstract Physics informed neural network (PINN) has become a promising method for solving partial differential equations (PDEs). The loss function of PINN is a weighted sum of multiple items. This makes it easy to fall into local optima, especially the gradient pathologies when solving high frequency problems. The value of penalty coefficients has a crucial impact on the prediction results. Therefore, a new PINN with adaptive penalty coefficients iteratively optimized by biased multiobjective coevolutionary algorithm (BC-PINN) is presented. In BC-PINN, a two-stage optimization mechanism is used to search for parameters of neural network and penalty coefficients respectively. This method involves constructing the fitness function of penalty coefficients based on the biased dominance ranking by data item and regularization item. Compared with the previous works of others, the accuracy of fitting the initial conditions and boundary conditions is considered to be given priority, which is more conducive to PINN converging to the particular solution of PDE. In addition, the set of penalty coefficients is divided into multiple populations to improve the optimization efficiency through coevolutionary algorithm. The empirical results show that: (1) Our method can improve the gradient pathologies and effectively capture the high-frequency features. (2) Compared to the original PINN, it reduces the MSE by 1–6 orders of magnitude in our benchmark functions.</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Physics informed neural network</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Gradient pathologies</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Coevolutionary algorithm</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Biased multiobjective optimization</subfield></datafield><datafield tag="700" ind1="1" ind2=" "><subfield code="a">Hao, Jia</subfield><subfield code="4">aut</subfield></datafield><datafield tag="700" ind1="1" ind2=" "><subfield code="a">Huang, Jin</subfield><subfield code="4">aut</subfield></datafield><datafield tag="700" ind1="1" ind2=" "><subfield code="a">Huang, Biao</subfield><subfield code="4">aut</subfield></datafield><datafield tag="773" ind1="0" ind2="8"><subfield code="i">Enthalten in</subfield><subfield code="t">Neural computing & applications</subfield><subfield code="d">Springer London, 1993</subfield><subfield code="g">35(2023), 28 vom: 02. Aug., Seite 21093-21113</subfield><subfield code="w">(DE-627)165669608</subfield><subfield code="w">(DE-600)1136944-9</subfield><subfield code="w">(DE-576)032873050</subfield><subfield code="x">0941-0643</subfield><subfield code="7">nnns</subfield></datafield><datafield tag="773" ind1="1" ind2="8"><subfield code="g">volume:35</subfield><subfield code="g">year:2023</subfield><subfield code="g">number:28</subfield><subfield code="g">day:02</subfield><subfield code="g">month:08</subfield><subfield code="g">pages:21093-21113</subfield></datafield><datafield tag="856" ind1="4" ind2="1"><subfield code="u">https://doi.org/10.1007/s00521-023-08876-4</subfield><subfield code="z">lizenzpflichtig</subfield><subfield code="3">Volltext</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_USEFLAG_A</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SYSFLAG_A</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_OLC</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SSG-OLC-MAT</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_2018</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4277</subfield></datafield><datafield tag="951" ind1=" " ind2=" "><subfield code="a">AR</subfield></datafield><datafield tag="952" ind1=" " ind2=" "><subfield code="d">35</subfield><subfield code="j">2023</subfield><subfield code="e">28</subfield><subfield code="b">02</subfield><subfield code="c">08</subfield><subfield code="h">21093-21113</subfield></datafield></record></collection>
|
author |
Zhu, Zhicheng |
spellingShingle |
Zhu, Zhicheng ddc 004 misc Physics informed neural network misc Gradient pathologies misc Coevolutionary algorithm misc Biased multiobjective optimization BC-PINN: an adaptive physics informed neural network based on biased multiobjective coevolutionary algorithm |
authorStr |
Zhu, Zhicheng |
ppnlink_with_tag_str_mv |
@@773@@(DE-627)165669608 |
format |
Article |
dewey-ones |
004 - Data processing & computer science |
delete_txt_mv |
keep |
author_role |
aut aut aut aut |
collection |
OLC |
remote_str |
false |
illustrated |
Not Illustrated |
issn |
0941-0643 |
topic_title |
004 VZ BC-PINN: an adaptive physics informed neural network based on biased multiobjective coevolutionary algorithm Physics informed neural network Gradient pathologies Coevolutionary algorithm Biased multiobjective optimization |
topic |
ddc 004 misc Physics informed neural network misc Gradient pathologies misc Coevolutionary algorithm misc Biased multiobjective optimization |
topic_unstemmed |
ddc 004 misc Physics informed neural network misc Gradient pathologies misc Coevolutionary algorithm misc Biased multiobjective optimization |
topic_browse |
ddc 004 misc Physics informed neural network misc Gradient pathologies misc Coevolutionary algorithm misc Biased multiobjective optimization |
format_facet |
Aufsätze Gedruckte Aufsätze |
format_main_str_mv |
Text Zeitschrift/Artikel |
carriertype_str_mv |
nc |
hierarchy_parent_title |
Neural computing & applications |
hierarchy_parent_id |
165669608 |
dewey-tens |
000 - Computer science, knowledge & systems |
hierarchy_top_title |
Neural computing & applications |
isfreeaccess_txt |
false |
familylinks_str_mv |
(DE-627)165669608 (DE-600)1136944-9 (DE-576)032873050 |
title |
BC-PINN: an adaptive physics informed neural network based on biased multiobjective coevolutionary algorithm |
ctrlnum |
(DE-627)OLC2145282378 (DE-He213)s00521-023-08876-4-p |
title_full |
BC-PINN: an adaptive physics informed neural network based on biased multiobjective coevolutionary algorithm |
author_sort |
Zhu, Zhicheng |
journal |
Neural computing & applications |
journalStr |
Neural computing & applications |
lang_code |
eng |
isOA_bool |
false |
dewey-hundreds |
000 - Computer science, information & general works |
recordtype |
marc |
publishDateSort |
2023 |
contenttype_str_mv |
txt |
container_start_page |
21093 |
author_browse |
Zhu, Zhicheng Hao, Jia Huang, Jin Huang, Biao |
container_volume |
35 |
class |
004 VZ |
format_se |
Aufsätze |
author-letter |
Zhu, Zhicheng |
doi_str_mv |
10.1007/s00521-023-08876-4 |
dewey-full |
004 |
title_sort |
bc-pinn: an adaptive physics informed neural network based on biased multiobjective coevolutionary algorithm |
title_auth |
BC-PINN: an adaptive physics informed neural network based on biased multiobjective coevolutionary algorithm |
abstract |
Abstract Physics informed neural network (PINN) has become a promising method for solving partial differential equations (PDEs). The loss function of PINN is a weighted sum of multiple items. This makes it easy to fall into local optima, especially the gradient pathologies when solving high frequency problems. The value of penalty coefficients has a crucial impact on the prediction results. Therefore, a new PINN with adaptive penalty coefficients iteratively optimized by biased multiobjective coevolutionary algorithm (BC-PINN) is presented. In BC-PINN, a two-stage optimization mechanism is used to search for parameters of neural network and penalty coefficients respectively. This method involves constructing the fitness function of penalty coefficients based on the biased dominance ranking by data item and regularization item. Compared with the previous works of others, the accuracy of fitting the initial conditions and boundary conditions is considered to be given priority, which is more conducive to PINN converging to the particular solution of PDE. In addition, the set of penalty coefficients is divided into multiple populations to improve the optimization efficiency through coevolutionary algorithm. The empirical results show that: (1) Our method can improve the gradient pathologies and effectively capture the high-frequency features. (2) Compared to the original PINN, it reduces the MSE by 1–6 orders of magnitude in our benchmark functions. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2023. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. |
abstractGer |
Abstract Physics informed neural network (PINN) has become a promising method for solving partial differential equations (PDEs). The loss function of PINN is a weighted sum of multiple items. This makes it easy to fall into local optima, especially the gradient pathologies when solving high frequency problems. The value of penalty coefficients has a crucial impact on the prediction results. Therefore, a new PINN with adaptive penalty coefficients iteratively optimized by biased multiobjective coevolutionary algorithm (BC-PINN) is presented. In BC-PINN, a two-stage optimization mechanism is used to search for parameters of neural network and penalty coefficients respectively. This method involves constructing the fitness function of penalty coefficients based on the biased dominance ranking by data item and regularization item. Compared with the previous works of others, the accuracy of fitting the initial conditions and boundary conditions is considered to be given priority, which is more conducive to PINN converging to the particular solution of PDE. In addition, the set of penalty coefficients is divided into multiple populations to improve the optimization efficiency through coevolutionary algorithm. The empirical results show that: (1) Our method can improve the gradient pathologies and effectively capture the high-frequency features. (2) Compared to the original PINN, it reduces the MSE by 1–6 orders of magnitude in our benchmark functions. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2023. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. |
abstract_unstemmed |
Abstract Physics informed neural network (PINN) has become a promising method for solving partial differential equations (PDEs). The loss function of PINN is a weighted sum of multiple items. This makes it easy to fall into local optima, especially the gradient pathologies when solving high frequency problems. The value of penalty coefficients has a crucial impact on the prediction results. Therefore, a new PINN with adaptive penalty coefficients iteratively optimized by biased multiobjective coevolutionary algorithm (BC-PINN) is presented. In BC-PINN, a two-stage optimization mechanism is used to search for parameters of neural network and penalty coefficients respectively. This method involves constructing the fitness function of penalty coefficients based on the biased dominance ranking by data item and regularization item. Compared with the previous works of others, the accuracy of fitting the initial conditions and boundary conditions is considered to be given priority, which is more conducive to PINN converging to the particular solution of PDE. In addition, the set of penalty coefficients is divided into multiple populations to improve the optimization efficiency through coevolutionary algorithm. The empirical results show that: (1) Our method can improve the gradient pathologies and effectively capture the high-frequency features. (2) Compared to the original PINN, it reduces the MSE by 1–6 orders of magnitude in our benchmark functions. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2023. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. |
collection_details |
GBV_USEFLAG_A SYSFLAG_A GBV_OLC SSG-OLC-MAT GBV_ILN_2018 GBV_ILN_4277 |
container_issue |
28 |
title_short |
BC-PINN: an adaptive physics informed neural network based on biased multiobjective coevolutionary algorithm |
url |
https://doi.org/10.1007/s00521-023-08876-4 |
remote_bool |
false |
author2 |
Hao, Jia Huang, Jin Huang, Biao |
author2Str |
Hao, Jia Huang, Jin Huang, Biao |
ppnlink |
165669608 |
mediatype_str_mv |
n |
isOA_txt |
false |
hochschulschrift_bool |
false |
doi_str |
10.1007/s00521-023-08876-4 |
up_date |
2024-07-04T02:36:24.827Z |
_version_ |
1803614247796604928 |
fullrecord_marcxml |
<?xml version="1.0" encoding="UTF-8"?><collection xmlns="http://www.loc.gov/MARC21/slim"><record><leader>01000naa a22002652 4500</leader><controlfield tag="001">OLC2145282378</controlfield><controlfield tag="003">DE-627</controlfield><controlfield tag="005">20240118103717.0</controlfield><controlfield tag="007">tu</controlfield><controlfield tag="008">240118s2023 xx ||||| 00| ||eng c</controlfield><datafield tag="024" ind1="7" ind2=" "><subfield code="a">10.1007/s00521-023-08876-4</subfield><subfield code="2">doi</subfield></datafield><datafield tag="035" ind1=" " ind2=" "><subfield code="a">(DE-627)OLC2145282378</subfield></datafield><datafield tag="035" ind1=" " ind2=" "><subfield code="a">(DE-He213)s00521-023-08876-4-p</subfield></datafield><datafield tag="040" ind1=" " ind2=" "><subfield code="a">DE-627</subfield><subfield code="b">ger</subfield><subfield code="c">DE-627</subfield><subfield code="e">rakwb</subfield></datafield><datafield tag="041" ind1=" " ind2=" "><subfield code="a">eng</subfield></datafield><datafield tag="082" ind1="0" ind2="4"><subfield code="a">004</subfield><subfield code="q">VZ</subfield></datafield><datafield tag="100" ind1="1" ind2=" "><subfield code="a">Zhu, Zhicheng</subfield><subfield code="e">verfasserin</subfield><subfield code="4">aut</subfield></datafield><datafield tag="245" ind1="1" ind2="0"><subfield code="a">BC-PINN: an adaptive physics informed neural network based on biased multiobjective coevolutionary algorithm</subfield></datafield><datafield tag="264" ind1=" " ind2="1"><subfield code="c">2023</subfield></datafield><datafield tag="336" ind1=" " ind2=" "><subfield code="a">Text</subfield><subfield code="b">txt</subfield><subfield code="2">rdacontent</subfield></datafield><datafield tag="337" ind1=" " ind2=" "><subfield code="a">ohne Hilfsmittel zu benutzen</subfield><subfield code="b">n</subfield><subfield code="2">rdamedia</subfield></datafield><datafield tag="338" ind1=" " ind2=" "><subfield code="a">Band</subfield><subfield code="b">nc</subfield><subfield code="2">rdacarrier</subfield></datafield><datafield tag="500" ind1=" " ind2=" "><subfield code="a">© The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2023. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.</subfield></datafield><datafield tag="520" ind1=" " ind2=" "><subfield code="a">Abstract Physics informed neural network (PINN) has become a promising method for solving partial differential equations (PDEs). The loss function of PINN is a weighted sum of multiple items. This makes it easy to fall into local optima, especially the gradient pathologies when solving high frequency problems. The value of penalty coefficients has a crucial impact on the prediction results. Therefore, a new PINN with adaptive penalty coefficients iteratively optimized by biased multiobjective coevolutionary algorithm (BC-PINN) is presented. In BC-PINN, a two-stage optimization mechanism is used to search for parameters of neural network and penalty coefficients respectively. This method involves constructing the fitness function of penalty coefficients based on the biased dominance ranking by data item and regularization item. Compared with the previous works of others, the accuracy of fitting the initial conditions and boundary conditions is considered to be given priority, which is more conducive to PINN converging to the particular solution of PDE. In addition, the set of penalty coefficients is divided into multiple populations to improve the optimization efficiency through coevolutionary algorithm. The empirical results show that: (1) Our method can improve the gradient pathologies and effectively capture the high-frequency features. (2) Compared to the original PINN, it reduces the MSE by 1–6 orders of magnitude in our benchmark functions.</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Physics informed neural network</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Gradient pathologies</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Coevolutionary algorithm</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Biased multiobjective optimization</subfield></datafield><datafield tag="700" ind1="1" ind2=" "><subfield code="a">Hao, Jia</subfield><subfield code="4">aut</subfield></datafield><datafield tag="700" ind1="1" ind2=" "><subfield code="a">Huang, Jin</subfield><subfield code="4">aut</subfield></datafield><datafield tag="700" ind1="1" ind2=" "><subfield code="a">Huang, Biao</subfield><subfield code="4">aut</subfield></datafield><datafield tag="773" ind1="0" ind2="8"><subfield code="i">Enthalten in</subfield><subfield code="t">Neural computing & applications</subfield><subfield code="d">Springer London, 1993</subfield><subfield code="g">35(2023), 28 vom: 02. Aug., Seite 21093-21113</subfield><subfield code="w">(DE-627)165669608</subfield><subfield code="w">(DE-600)1136944-9</subfield><subfield code="w">(DE-576)032873050</subfield><subfield code="x">0941-0643</subfield><subfield code="7">nnns</subfield></datafield><datafield tag="773" ind1="1" ind2="8"><subfield code="g">volume:35</subfield><subfield code="g">year:2023</subfield><subfield code="g">number:28</subfield><subfield code="g">day:02</subfield><subfield code="g">month:08</subfield><subfield code="g">pages:21093-21113</subfield></datafield><datafield tag="856" ind1="4" ind2="1"><subfield code="u">https://doi.org/10.1007/s00521-023-08876-4</subfield><subfield code="z">lizenzpflichtig</subfield><subfield code="3">Volltext</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_USEFLAG_A</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SYSFLAG_A</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_OLC</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SSG-OLC-MAT</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_2018</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4277</subfield></datafield><datafield tag="951" ind1=" " ind2=" "><subfield code="a">AR</subfield></datafield><datafield tag="952" ind1=" " ind2=" "><subfield code="d">35</subfield><subfield code="j">2023</subfield><subfield code="e">28</subfield><subfield code="b">02</subfield><subfield code="c">08</subfield><subfield code="h">21093-21113</subfield></datafield></record></collection>
|
score |
7.4001493 |