Dr. Jawad Fayaz, acknowledges the Department of Computer Science at the University of Exeter, UK, for financial support provided for the high-performance computing. The second author, Dr
Rodrigo Astroza, acknowledges the financial support from the Chilean National Research and Development Agency (Agencia Nacional de Investigación y Desarrollo, ANID) through Fondo Nacional de Desarrollo Científico y Tecnológico (FONDECYT) Regular 1240503 and Fondo de Valorización de la Investigación (FOVI)230030 projects. The third author, Dr
Sergio Ruiz, acknowledges the financial support from the ANID through FONDECYT Regular 1240501.
In the face of the unrelenting challenge posed by earthquakes—a natural hazard of unpredictable nature with a legacy of significant loss of life, destruction of infrastructure, and profound economic and social impacts—the scientific community has pursued advancements in earthquake early warning systems (EEWSs). These systems are vital for pre-emptive actions and decision-making that can save lives and safeguard critical infrastructure. This study proposes and validates a domain-informed deep learning-based EEWS called the hybrid earthquake early warning framework for estimating response spectra (HEWFERS), which represents a significant leap forward in the capabilities to predict ground shaking intensity in real-time, aligning with the United Nations’ disaster risk reduction goals. HEWFERS ingeniously integrates a domain-informed variational autoencoder for physics-based latent variable (LV) extraction, a feed-forward neural network for on-site prediction, and Gaussian process regression for spatial prediction. Adopting explainable artificial intelligence-based Shapley explanations further elucidates the predictive mechanisms, ensuring stakeholder-informed decisions. By conducting an extensive analysis of the proposed framework under a large database of approximately 14 000 recorded ground motions, this study offers insights into the potential of integrating machine learning with seismology to revolutionize earthquake preparedness and response, thus paving the way for a safer and more resilient future.
Dr. Jawad Fayaz, acknowledges the Department of Computer Science at the University of Exeter, UK, for financial support provided for the high-performance computing. The second author, Dr
"}, bioImg=null, bioContent=
Dr. Jawad Fayaz, acknowledges the Department of Computer Science at the University of Exeter, UK, for financial support provided for the high-performance computing. The second author, Dr
, aboutCorrespAuthor=null), CN=AuthorExt(id=1249026618799346672, tenantId=1045748351789510663, journalId=1155139928190095384, articleId=1159996682937098495, authorId=1249026618677711849, language=CN, stringName=Jawad Fayaz, firstName=Jawad, middleName=null, lastName=Fayaz, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=a, b, address=a.Department of Computer Science, University of Exeter, Exeter, EX4 4RN, UK b.Department of Civil, Environmental, and Geomatic Engineering, University College London, London, WC1E 6BT, UK, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1249026618434442196, tenantId=1045748351789510663, journalId=1155139928190095384, articleId=1159996682937098495, xref=a., ext=[AuthorCompanyExt(id=1249026618451219414, tenantId=1045748351789510663, journalId=1155139928190095384, articleId=1159996682937098495, companyId=1249026618434442196, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=a.Department of Computer Science, University of Exeter, Exeter, EX4 4RN, UK)]), AuthorCompany(id=1249026618493162457, tenantId=1045748351789510663, journalId=1155139928190095384, articleId=1159996682937098495, xref=b., ext=[AuthorCompanyExt(id=1249026618509939674, tenantId=1045748351789510663, journalId=1155139928190095384, articleId=1159996682937098495, companyId=1249026618493162457, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=b.Department of Civil, Environmental, and Geomatic Engineering, University College London, London, WC1E 6BT, UK)])]), Author(id=1249026618845484020, tenantId=1045748351789510663, journalId=1155139928190095384, articleId=1159996682937098495, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=rastroza@miuandes.cl, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1249026618908398583, tenantId=1045748351789510663, journalId=1155139928190095384, articleId=1159996682937098495, authorId=1249026618845484020, language=EN, stringName=null, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=c, address=null, bio={"content":"
Rodrigo Astroza, acknowledges the financial support from the Chilean National Research and Development Agency (Agencia Nacional de Investigación y Desarrollo, ANID) through Fondo Nacional de Desarrollo Científico y Tecnológico (FONDECYT) Regular 1240503 and Fondo de Valorización de la Investigación (FOVI)230030 projects. The third author, Dr
"}, bioImg=null, bioContent=
Rodrigo Astroza, acknowledges the financial support from the Chilean National Research and Development Agency (Agencia Nacional de Investigación y Desarrollo, ANID) through Fondo Nacional de Desarrollo Científico y Tecnológico (FONDECYT) Regular 1240503 and Fondo de Valorización de la Investigación (FOVI)230030 projects. The third author, Dr
, aboutCorrespAuthor=null), CN=AuthorExt(id=1249026618954535929, tenantId=1045748351789510663, journalId=1155139928190095384, articleId=1159996682937098495, authorId=1249026618845484020, language=CN, stringName=Rodrigo Astroza, firstName=Rodrigo, middleName=null, lastName=Astroza, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=c, address=c.Facultad de Ingeniería y Ciencias Aplicadas, Universidad de los Andes, Santiago 7620001, Chile, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1249026618556077021, tenantId=1045748351789510663, journalId=1155139928190095384, articleId=1159996682937098495, xref=c., ext=[AuthorCompanyExt(id=1249026618568659934, tenantId=1045748351789510663, journalId=1155139928190095384, articleId=1159996682937098495, companyId=1249026618556077021, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=c.Facultad de Ingeniería y Ciencias Aplicadas, Universidad de los Andes, Santiago 7620001, Chile)])]), Author(id=1249026619004867580, tenantId=1045748351789510663, journalId=1155139928190095384, articleId=1159996682937098495, orderNo=2, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1249026619063587839, tenantId=1045748351789510663, journalId=1155139928190095384, articleId=1159996682937098495, authorId=1249026619004867580, language=EN, stringName=null, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=d, address=null, bio={"content":"
Sergio Ruiz, acknowledges the financial support from the ANID through FONDECYT Regular 1240501.
"}, bioImg=null, bioContent=
Sergio Ruiz, acknowledges the financial support from the ANID through FONDECYT Regular 1240501.
, aboutCorrespAuthor=null), CN=AuthorExt(id=1249026619109724160, tenantId=1045748351789510663, journalId=1155139928190095384, articleId=1159996682937098495, authorId=1249026619004867580, language=CN, stringName=Sergio Ruiz, firstName=Sergio, middleName=null, lastName=Ruiz, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=d, address=d.Department of Geophysics, University of Chile, Santiago 8330111, Chile, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1249026618618991586, tenantId=1045748351789510663, journalId=1155139928190095384, articleId=1159996682937098495, xref=d., ext=[AuthorCompanyExt(id=1249026618631574500, tenantId=1045748351789510663, journalId=1155139928190095384, articleId=1159996682937098495, companyId=1249026618618991586, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=d.Department of Geophysics, University of Chile, Santiago 8330111, Chile)])])]
Jawad Fayaz,Rodrigo Astroza,Sergio Ruiz.
可解释且基于领域感知的实时混合地震早期预警系统——地震动强度预测[J].
工程(英文), 2025, 49(6): 190-204 DOI:10.1016/j.eng.2025.03.009
图3展示了领域感知VAE的结构,其中, x 表示输入变量的真实向量[本研究中为85个点的Sa(T)谱], x̂ 表示输入变量的重构向量[本研究中为预测的85个点的Sa(T)谱]。VAE通过基于神经网络的编码器(识别模型),通过概率推测,将向量形式的观测数据映射到LV空间。而该编码器与基于神经网络的解码器(生成模型)共同训练,解码器利用LV空间对观测数据进行重构。因此,将LV空间设计为连续平滑的表示形式,使得在潜空间中数值相近的点经解码器重构后,对应的结果也相似。
VAE训练背后的核心概念基于贝叶斯定理[如公式(1)所示],其中, z 表示 LV 空间,p( z )为先验分布,p( x|z )为给定 z 时 x 的条件概率分布。由于 x 的概率密度p( x )难以直接计算,可通过变分推断[59]进行近似,该方法通过估计给定 x 时 z 的条件概率分布p( z|x ),用另一个可处理的近似分布q( z|x )来替代,实现对p( x )的近似估计。任意两个随机变量y的概率分布p(y)和q(y)之间的KL散度(Kullback-Leibler divergence)(记为KL[q|p])可通过公式(2)计算得到。通过设计近似分布q( z|x )使其尽可能接近真实分布p( z|x ),即可对难以直接求解的分布进行近似推断。这一过程通过最小化公式(3)中给出的q( z|x )与p( z|x )之间的KL散度损失(记为LossKL)[68]来实现。在此背景下,p( z|x )采用先验分布p( z ),假设其服从单位高斯分布[N(0,1)],其中每个LV均服从该分布,N为批次大小。公式(4)代表重构损失(Lossrecon),q(z|x) log p( x|z )对应真实值 x 与预测值 x̂ 之间的均方误差(mean squared error, MSE),代表期望函数[59]。Lossrecon用于减少VAE在真实值与预测值之间的偏差,LossKL则有助于生成紧凑且连续的潜在表示。
RitchieH, RosadoP, RoserM. Natural disasters [Internet]. Oxford: Our World in Data; [cited 2025 Feb 18]. Available from:
[2]
SwissR. Global natural and manmade disasters cost insurers US$83B in 2020 [Internet]. San Diego: Wells Media Group, Inc.; 2020 Dec 16 [cited 2025 Feb 18]. Available from:
[3]
XuS, DimasakaJ, WaldDJ, NohHY. Seismic multi-hazard and impact estimation via causal inference from satellite imagery. Nat Commun2022;13(1):7793. . 10.1038/s41467-022-35418-8
[4]
BataMH, CarriveauR, TingDSK. Urban water supply systems’ resilience under earthquake scenario. Sci Rep2022;12(1):20555. . 10.1038/s41598-022-23126-8
[5]
ZhangS, ZhangL, LacasseS, NadimF. Evolution of mass movements near epicentre of Wenchuan earthquake, the first eight years. Sci Rep2016;6(1):36154. . 10.1038/srep36154
[6]
LeónJA, OrdazM, HaddadE, AraújoIF. Risk caused by the propagation of earthquake losses through the economy. Nat Commun2022;13(1):2908. . 10.1038/s41467-022-30504-3
[7]
YunusAP, XinyuC, CataniF, SubramaniamSS, FanX, JieD, et al. Earthquake induced soil landslides: volume estimates and uncertainties with the existing scaling exponents. Sci Rep2023;13(1):8151. . 10.1038/s41598-023-35088-6
[8]
Guerrero-MirandaP, LuqueGA. Social responsibility, sustainability, and public policy: the lessons of debris management after the manabí earthquake in ecuador. Int J Environ Res Public Health2021;18(7):3494. . 10.3390/IJERPH18073494
[9]
MavroulisS, MavrouliM, VassilakisE, ArgyropoulosI, CarydisP, LekkasE. Debris management in turkey provinces affected by the 6 February 2023 earthquakes: challenges during recovery and potential health and environmental risks. Appl Sci2023;13(15):8823. . 10.3390/app13158823
[10]
MavroulisS, MavrouliM, LekkasE, TsakrisA. Managing earthquake debris: environmental issues, health impacts, and risk reduction measures. Environments2023;10(11):192. . 10.3390/environments10110192
ChungAI. The development of earthquake early warning methods. Nat Rev Earth Environ2020;1(7):331. . 10.1038/s43017-020-0070-x
[13]
CremenG, GalassoC, ZuccoloE. Investigating the potential effectiveness of earthquake early warning across Europe. Nat Commun2022;13(1):639. . 10.1038/s41467-021-27807-2
[14]
AllenRM, StogaitisM. Global growth of earthquake early warning. Science2022;375(6582):717‒8. . 10.1126/science.abl5435
[15]
United Nations Office for Disaster Risk Reduction (UNDRR). Sendai framework for disaster risk reduction 2015‒2030. Report. Boon: United Nations Office for Disaster Risk Reduction; 2022.
[16]
GalassoC, OpabolaEA. The 2023 Kahramanmaras earthquake sequence: f inding a path to a more resilient, sustainable, and equitable society. Commun Eng2024;3(1):24. . 10.1038/s44172-024-00170-y
[17]
KajitaniY, ChangSE, TatanoH. Economic impacts of the 2011 Tohoku-Oki earthquake and tsunami. Earthq Spectra2013;29():457‒78. . 10.1193/1.4000108
CenterCopenhagen Consensus. Natural disasters [Internet]. Copenhagen: Copenhagen Consensus Center; [cited 2025 Feb 18]. Available from:10.2307/j.ctv35r4b5g
[20]
KoderaY. An earthquake early warning method based on Huygens principle: robust ground motion prediction using various localized distance-attenuation models. J Geophys Res Solid Earth2019;124(12):12981‒96. . 10.1029/2019jb017862
[21]
HoshibaM. Real-time prediction of ground motion by Kirchhoff-Fresnel boundary integral equation method: extended front detection method for earthquake early warning. J Geophys Res Solid Earth2013;118(3):1038‒50. . 10.1002/jgrb.50119
[22]
GrapenthinR, JohansonIA, AllenRM. Operational real-time GPS-enhanced earthquake early warning. J Geophys Res Solid Earth2014;119(10):7944‒65. . 10.1002/2014jb011400
[23]
ChungAI, HensonI, AllenRM. Optimizing earthquake early warning performance: ElarmS-3. Seismol Res Lett2019;90(2A):727‒43. . 10.1785/0220180192
[24]
SuzukiW, AoiS, SekiguchiH, KunugiT. Rupture process of the 2011 Tohoku Oki mega-thrust earthquake (M 9.0) inverted from strong-motion data. Geophys Res Lett 2011;38(7):2011GL049136. . 10.1029/2011gl049136
[25]
ZhangJ, ZhangH, ChenE, ZhengY, KuangW, ZhangX. Real-time earthquake monitoring using a search engine method. Nat Commun2014;5(1):5664. . 10.1038/ncomms6664
KuangW, YuanC, ZhangJ. Real-time determination of earthquake focal mechanism via deep learning. Nat Commun2021;12(1):1432. . 10.1038/s41467-021-21670-x
[28]
CremenG, GalassoC. Earthquake early warning: recent advances and perspectives. Earth Sci Rev2020;205:103184. . 10.1016/j.earscirev.2020.103184
[29]
IaccarinoAG, PicozziM, BindiD, SpallarossaD. On-site earthquake early warning: predictive models for acceleration response spectra considering site effects. Bull Seismol Soc Am2020;110:1289‒304. . 10.1785/0120190272
[30]
BhardwajR, SharmaML, KumarA. Multi-parameter algorithm for earthquake early warning. Geomatics Nat Hazards Risk2016;7(4):1242‒64. . 10.1080/19475705.2015.1069409
[31]
CarusoA, ColombelliS, EliaL, PicozziM, ZolloA. An on-site alert level early warning system for Italy. J Geophys Res Solid Earth2017;122(3):2106‒18. . 10.1002/2016jb013403
[32]
LinCCJ, LinPY, ChangTM, LinTK. Development of on-site earthquake early warning system for Taiwan. In: D’Amico S, editor. Earthquake research and analysis—new frontiers in seismology. London: IntechOpen Ltd.; 2012. p. 329‒58. . 10.5772/28056
[33]
MeierM, KoderaY, BöseM, ChungA, HoshibaM, CochranE, et al. How often can earthquake early warning systems alert sites with high-intensity ground motion? J Geophys Res Solid Earth2020;125(2):1‒17. . 10.1029/2019jb017718
[34]
MünchmeyerJ, BindiD, LeserU, TilmannF. The transformer earthquake alerting model: a new versatile approach to earthquake early warning. Geophys J Int2021;225(1):646‒56. . 10.1093/gji/ggaa609
[35]
WuS, BeckJL, HeatonTH. ePAD: earthquake probability-based automated decision-making framework for earthquake early warning. Comput-Aided Civ Infrastruct Eng2013;28(10):737‒52. . 10.1111/mice.12048
[36]
FayazJ, GalassoC. A deep neural network framework for real-time on-site estimation of acceleration response spectra of seismic ground motions. Comput Aided Civ Infrastruct Eng2023;38(1):87‒103. . 10.1111/mice.12830
[37]
TajimaF, HayashidaT. Earthquake early warning: what does “seconds before a strong hit” mean? Prog Earth Planet Sci2018;5(1):63. . 10.1186/s40645-018-0221-6
[38]
WaldDJ. Practical limitations of earthquake early warning. Earthq Spectra2020;36(3):1412‒47. . 10.1177/8755293020911388
[39]
MinsonSE, MeierMA, BaltayAS, HanksTC, CochranES. The limits of earthquake early warning: timeliness of ground motion estimates. Sci Adv 2018;4(3):eaaq0504. . 10.1126/sciadv.aaq0504
[40]
MinsonSE, BaltayAS, CochranES, HanksTC, PageMT, McBrideSK, et al. The limits of earthquake early warning accuracy and best alerting strategy. Sci Rep2019;9(1):2478. . 10.1038/s41598-019-39384-y
[41]
MittalH, YangBM, WuYM. Progress on the earthquake early warning and shakemaps system using low-cost sensors in Taiwan. Geosci Lett2022;9(1):42. . 10.1186/s40562-022-00251-w
[42]
ChiangYJ, ChinTL, ChenDY. Neural network-based strong motion prediction for on-site earthquake early warning. Sensors2022;22(3):704. . 10.3390/s22030704
[43]
CampbellKW, BozorgniaY. NGA-West2 ground motion model for the average horizontal components of PGA, PGV, and 5% damped linear acceleration response spectra. Earthq Spectra2014;30(3):1087‒115. . 10.1193/062913eqs175m
[44]
FayazJ, XiangY, ZareianF. Generalized ground motion prediction model using hybrid recurrent neural network. Earthq Eng Struct Dyn2021;50(6):1539‒61. . 10.1002/eqe.3410
[45]
AbrahamsonNA, SilvaWJ, KamaiR. Summary of the ASK14 ground motion relation for active crustal regions. Earthq Spectra2014;30(3):1025‒55. . 10.1193/070913eqs198m
[46]
McBrideSK, BostromA, SuttonJ, de GrootRM, BaltayAS, TerbushB, et al. Developing post-alert messaging for ShakeAlert, the earthquake early warning system for the West Coast of the United States of America. Int J Disaster Risk Reduct2020;50:101713. . 10.1016/j.ijdrr.2020.101713
[47]
BakerJW. Efficient analytical fragility function fitting using dynamic structural analysis. Earthq Spectra2015;31(1):579‒99. . 10.1193/021113eqs025m
[48]
BakerJW, AllinCC. A vector-valued ground motion intensity measure consisting of spectral acceleration and epsilon. Earthq Eng Struct Dyn2005;34(10):1193‒217. . 10.1002/eqe.474
[49]
JalayerF, BeckJL, ZareianF. Analyzing the sufficiency of alternative scalar and vector intensity measures of ground shaking based on information theory. J Eng Mech2012;138(3):307‒16. . 10.1061/(asce)em.1943-7889.0000327
BurattiN, StaffordPJ, BommerJJ. Earthquake accelerogram selection and scaling procedures for estimating the distribution of drift response. J Struct Eng2011;137(3):345‒57. . 10.1061/(asce)st.1943-541x.0000217
[52]
TothongP, LucoN. Probabilistic seismic demand analysis using advanced ground motion intensity measures. Earthq Eng Struct Dyn2007;36(13):1837‒60. . 10.1002/eqe.696
[53]
BazzurroP, CornellCA, ShomeN, CarballoJE. Three proposals for characterizing MDOF nonlinear seismic response. J Struct Eng1998;124(11):1281‒9. . 10.1061/(asce)0733-9445(1998)124:11(1281)
[54]
FayazJ, ZareianF. An efficient algorithm to simulate site-based ground motions that match a target spectrum. Earthq Eng Struct Dyn2021;50(13):3532‒49. . 10.1002/eqe.3521
[55]
FayazJ, AzarS, DabaghiM, ZareianF. Methodology for validation of simulated ground motions for seismic response assessment: application to cybershake source-based ground motions. Bull Seismol Soc Am2021;111(1):226‒41. . 10.1785/0120200240
[56]
FayazJ, AzarS, DabaghiM, ZareianF. An efficient algorithm to simulate hazard-targeted site-based synthetic ground motions. Earthq Spectra2021;37(2):876‒902. . 10.1177/8755293020970968
KohrangiM, VamvatsikosD, BazzurroP. Site dependence and record selection schemes for building fragility and regional loss assessment. Earthq Eng Struct Dyn2017;46(10):1625‒43. . 10.1002/eqe.2873
[59]
KingmaDP, WellingM. An introduction to variational autoencoders. Found Trends Mach Learn2019;12(4):307‒92. . 10.1561/2200000056
[60]
RosenblattF. The perceptron: a perceiving and recognizing automaton. Report. Amherst: University of Massachusetts Amherst; 1957.
[61]
WilliamsCKI, RasmussenCE. Gaussian Processes for Regression. In: Proceedings of the 9th International Conference on Neural Information Processing Systems; 1995 Nov 27‒Dec 2; Denver, CO, USA. Cambridge: MIT Press; 1995. p. 514‒20. . 10.1109/ijcnn.2016.7727626
[62]
HolzerTL, SavageJC. Global earthquake fatalities and population. Earthq Spectra2013;29(1):155‒75. . 10.1193/1.4000106
[63]
National Research Institute for Earth Science and Disaster Resilience. Strong motion seismograph networks (NIED K-NET, KiK-Net) [Internet]. Tokyo: National Research Institute for Earth Science and Disaster Resilience; [cited 2025 Feb 18]. Available from:
[64]
HsuTY, WuRT, ChangKC. Two novel approaches to reduce false alarm due to non-earthquake events for on-site earthquake early warning system. Comput Aided Civ Infrastruct Eng2016;31(7):535‒49. . 10.1111/mice.12191
AkazawaT. A technique for automatic detection of onset time of p-and s phases in strong motion records. In: Proceedings of the 15th World Conference on Earthquake Engineering (15WCEE); 2012 Sep 24‒28; Lisbon, Portugal. Tokyo: International Association for Earthquake Engineering (IAEE); 2004.
[67]
SleemanR, van EckT. Robust automatic p-phase picking: an on-line implementation in the analysis of broadband seismogram recordings. Phys Earth Planet Inter 1999;113(1‒4):265‒75.
[68]
KullbackS, LeiblerRA. On information and sufficiency. Ann Math Stat1951;22(1):79‒86. . 10.1214/aoms/1177729694
[69]
BergstraJ, BengioY. Random search for hyper-parameter optimization. J Mach Learn Res2012;13:281‒305.
[70]
CampbellKW, BozorgniaY. Ground motion models for the horizontal components of Arias intensity (AI) and cumulative absolute velocity (CAV) using the NGA-West2 database. Earthq Spectra2019;35(3):1289‒310. . 10.1193/090818eqs212m
[71]
SandıkkayaMA, AkkarS. Cumulative absolute velocity, Arias intensity and significant duration predictive models from a pan-European strong-motion dataset. Bull Earthq Eng2017;15(5):1881‒98. . 10.1007/s10518-016-0066-6
[72]
WuYM, KanamoriH, AllenRM, HaukssonE. Determination of earthquake early warning parameters, τc and Pd, for Southern California. Geophys J Int2007;170(2):711‒7. . 10.1111/j.1365-246x.2007.03430.x
[73]
FayazJ, MedallaM, ZareianF. Sensitivity of the response of box-girder seat type bridges to the duration of ground motions arising from crustal and subduction earthquakes. Eng Struct2020;219:110845. . 10.1016/j.engstruct.2020.110845
[74]
FayazJ, MedallaM, Torres-RodasP, GalassoC. A recurrent-neural-network based generalized ground-motion model for the Chilean subduction seismic environment. Struct Saf2023;100:102282. . 10.1016/j.strusafe.2022.102282
ChenT, GuestrinC. XGBoost: a scalable tree boosting system. In: Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining; 2016 Aug 13‒17; FranciscoSan, CA, USA. New York City: Association for Computing Machinery (ACM); 2016. p. 785‒94. . 10.1145/2939672.2939785
[77]
DemidenkoE. Mixed models: theory and applications with R. 2nd ed. Hoboken: Wiley Publisher; 2013.
[78]
NealRM. Bayesian learning for neural networks. New York City: Springer New York; 1996. . 10.1007/978-1-4612-0745-0
KroeseDP, TaimreT, BotevZI. Handbook of Monte Carlo methods. New York City: John Wiley and Sons; 2011. . 10.1002/9781118014967
[81]
VatsD. Flegal JM, Jones GL. Multivariate output analysis for Markov chain Monte Carlo. 2015. arXiv:
[82]
GelmanA, RubinD. Inference from iterative simulation using multiple sequences. Stat Sci1992;7(4):457‒511. . 10.1214/ss/1177011136
[83]
WangM, WangG. Damping-dependent correlations between horizontal-horizontal, horizontal-vertical, and vertical-vertical pairs of spectral accelerations. Earthq Eng Struct Dyn2024;53(7):2423‒44. . 10.1002/eqe.4119
[84]
LundbergSM, LeeSI. A unified approach to interpreting model predictions. In: Proceedings of the 31st International Conference on Neural Information Processing Systems; 2017 Dec 4‒9; LongBeach, CA, USA. Red Hook: Curran Associates Inc.; 2017. p. 4768‒77.
[85]
RothAE. The Shapley value: essays in honor of Lloyd S. Shapley. Cambridge: Cambridge University Press; 1988. . 10.1017/CBO9780511528446.015
[86]
MolnarC. Interpretable machine learning—a guide for making black box models explainable. San Francisco: Lulu Press; 2020.
[87]
BozorgniaY, BerteroVV. Earthquake engineering: from engineering seismology to performance-based engineering. Boca Raton: CRC Press; 2004.