Multi-task deep transfer learning for complicated seismic dynamic response prediction in slope systems

Abstract

Slope engineering is an uncertain, dynamic, and complex nonlinear spatiotemporal system with time delays. High-fidelity prediction of slope seismic stability has long been a formidable challenge due to the inherent randomness and uncertainty associated with ground motion, geo-material properties, complex topography, etc. Traditional numerical modelling always takes a simplified model by forcedly ignoring those uncertainties, thus failing to replicate precisely the intricate nonlinear interactions between factors that affect slope instability. Notably, the newly emerging deep learning methods have the capability of handling multiple factors with uncertainties. However, these methods heavily rely on extensive and comprehensive sensor data, while arranging sensors at certain important positions is sometimes unachievable. Therefore, we propose a multi-task deep transfer learning (MT-DTL) framework in this study to enhance the prediction accuracy of slope seismic response especially in data-limited conditions. The dynamic response at the locations without sufficient accessible sensor data can be effectively predicted with a newly developed algorithm. To collect the necessary sensor data, we conduct a series of physics experiments with the world’s largest multifunctional shaking table equipment. We demonstrate the efficacy and accuracy of our approach on the shaking-table datasets through comparisons with traditional machine learning (ML) methods. Our findings reveal that the MT-DTL framework can improve the confidence level of prediction results (within 5%) from the highest 86.4% by the optimal traditional ML methods to 92.7%, achieving comparable results with two-thirds fewer data. Additionally, a single response example showed that the trained deep transfer learning model has significantly improved the computational efficiency (0.018 - 0.019 s) compared to the dynamic finite element calculation with GeoStudio (10 min). This highlights its potential for integration into geo-hazards digital twin systems, facilitating rapid risk analysis based on real-time monitoring data.

Keywords

Slope seismic stability / Dynamic response / Deep transfer learning / Multi-task learning

CRediT authorship contribution statement

Xu Han: Writing - original draft, Validation, Software, Methodology, Data curation, Conceptualization. Yu Huang: Writing - review & editing, Supervision, Resources, Project administration, Funding acquisition, Conceptualization. Xiaoyan Jin: Writing - original draft, Visualization, Formal analysis. Liuyuan Zhao: Validation, Investigation, Data curation. Chung Yee Kwok: Writing - review & editing, Supervision.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

We thank the support offered by the National Natural Science Foundation of China (Grant No. 42120104008) and the Hong Kong Ph.D. Fellowship Scheme.

References

[1] Allstadt, K.E., Jibson, R.W., Thompson, E.M., Massey, C.I., Wald, D.J., Godt, J.W., Rengers, F.K., 2018. Improving near-real-time coseismic landslide models: lessons learned from the 2016 Kaikoura, New Zealand, earthquake. Bull. Seismol. Soc. Am. 108 (3B), 1649-1664.

[2] Banimelhem, O., Al-Khateeb, B., 2024. Facial image recognition using transfer learning: A brief review. 15th IEEE International Conference on Information and Communication Systems (ICICS), Irbid, Jordan, pp. 1-6.

[3] Bradley, K., Mallick, R., Andikagumi, H., Hubbard, J., Meilianda, E., Switzer, A., Du, N.R., Brocard, G., Alfian, D., Benazir, B., Feng, G.C., Yun, S.H., Majewski, J., Wei, S.J., Hill, E.M., 2019. Earthquake-triggered 2018 Palu Valley landslides enabled by wet rice cultivation. Nat. Geosci. 12, 935-939.

[4] Cacciola, P., Deodatis, G., 2011. A method for generating fully non-stationary and spectrum-compatible ground motion vector processes. Soil Dyn. Earthq. Eng. 31 (3), 351-360.

[5] Cai, T.H., Gan, H.Y., Peng, B., Huang, Q.Y., Zou, Z.Q., 2022. Real-time classification of disaster images from social media with a self-supervised learning framework. In: IEEE International Geoscience and Remote Sensing Symposium (IGARSS), Kuala Lumpur, pp. 671-674.

[6] Cho, Y.K., Khosravikia, F., Rathje, E.M., 2022. A comparison of artificial neural network and classical regression models for earthquake-induced slope displacements. Soil Dyn. Earthq. Eng. 152, 107024.

[7] Devi, A.C., Maragatham, S., Santhi, R., Srivastava, S., Dey, P., Barathkumar, S., 2025. Multivariate time series models for soil nutrient and yield prediction in site specific integrated nutrient management. Commun. Soil Sci. Plant Anal. 56 (12), 1889-1907.

[8] Du, P., Li, L.L., Kopf, A., Wang, D.W., Chen, K.J., Shi, H.B., Wang, W.T., Pan, X.Y., Hu, G., Zhang, P.Z., 2025. Earthquake-induced submarine landslides (EQISLs) and a comparison with their terrestrial counterparts: Insights from a new database. Earth-Sci. Rev. 261, 105021.

[9] Fan, X., Wang, X., Fang, C., Jansen, J.D., Dai, L., Tanyas, H., Zang, N., Tang, R., Xu, Q., Huang, R., 2025. Deep learning can predict global earthquake-triggered landslides. Natl. Sci. Rev. 12 (7), nwaf179. https://doi.org/10.1093/nsr/nwaf179.

[10] Fan, X.M., Scaringi, G., Korup, O., West, A.J., van Westen, C.J., Tanyas, H., Hovius, N., Hales, T.C., Jibson, R.W., Allstadt, K.E., Zhang, L.M., Evans, S.G., Xu, C., Li, G., Pei, X.J., Xu, Q., Huang, R.Q., 2019. Earthquake-induced chains of geologic hazards: Patterns, mechanisms, and impacts. Rev. Geophys. 57, 421-503.

[11] Fang, S.S., Yuan, Y., Chen, L., Zhang, J., Li, M., Song, S.D., 2017. Crop disease image recognition based on transfer learning. In: Proc. 9th International Conference on Image and Graphics (ICIG), Shanghai, China, pp. 545-554.

[12] Fang, Z.C., Wang, Y., van Westen, C., Lombardo, L., 2024. Landslide hazard spatiotemporal prediction based on data-driven models: Estimating where, when and how large landslide may be. Int. J. Appl. Earth Obs. Geoinf. 126, 103631.

[13] Frith, N.V., Hilton, R.G., Howarth, J.D., Gröcke, D.R., Fitzsimons, S.J., Croissant, T., Wang, J., McClymont, E.L., Dahl, J., Densmore, A.L., 2018. Carbon export from mountain forests enhanced by earthquake-triggered landslides over millennia. Nat. Geosci. 11 (10), 772-776.

[14] Galindo-Aires, R., Lara-Galera, A., Melentijevic, S., 2019. Hysteresis model for dynamic load under large strains. Int. J. Geomech. 19 (6), 04019051.

[15] Gischig, V.S., Eberhardt, E., Moore, J.R., Hungr, O., 2015. On the seismic response of deep-seated rock slope instabilities-Insights from numerical modeling. Eng. Geol. 193, 1-18.

[16] He, Q., Wang, M., Liu, K., 2021. Rapidly assessing earthquake-induced landslide susceptibility on a global scale using random forest. Geomorphology 391, 107889.

[17] Heidenreich, J.N., Bonatti, C., Mohr, D., 2024. Transfer learning of recurrent neural network-based plasticity models. Int. J. Numer. Methods Eng. 125 (1), e7357.

[18] Hu, H., Bao, Y., Huang, Y., Xiong, M., Wang, W., 2025. PDET-based assumption-free method for efficient seismic fragility assessment of slopes with combined uncertainties of soil materials and input ground motions. Reliab. Eng. Syst. Saf. 261, 111132.

[19] Huang, Y., Han, X., Zhao, L.Y., 2021. Recurrent neural networks for complicated seismic dynamic response prediction of a slope system. Eng. Geol. 289, 106198.

[20] Huang, Y., Hu, H.Q., Xiong, M., 2018. Probability density evolution method for seismic displacement-based assessment of earth retaining structures. Eng. Geol. 234, 167-173.

[21] Hwang, Y.W., Rathje, E.M., 2023. Insights into seismic slope deformation patterns using finite element analysis. Soil Dyn. Earthq. Eng. 164, 107660.

[22] Jibson, R.W., 2011. Methods for assessing the stability of slopes during earthquakes- A retrospective. Eng. Geol. 122 (1-2), 43-50.

[23] Kramer, S.L., Stewart, J.P., 2024. Geotechnical Earthquake Engineering. CRC Press, Taylor & Francis, p. 1060.

[24] Liao, W.W., Ji, J., Zhang, T., Wu, Z.J., Zhang, J., 2022. Time-dependent hazard assessment of rainfall-induced shallow landslides considering the spatial variability of soil permeability. Rock Soil Mech. 43 (S1), 623-632.

[25] Liu, C., Ge, L.L., Sepasgozar, S.M.E., 2021. Post-disaster classification of building damage using transfer learning. In: IEEE International Geoscience and Remote Sensing Symposium (IGARSS), Brussels, Belgium, pp. 2194-2197.

[26] Liu, F.Y., Li, J.L., Wang, L.B., 2023. PI-LSTM: Physics-informed long short-term memory network for structural response modeling. Eng. Struct. 292, 116500.

[27] Liu, R.L., Han, Y.H., Xiao, J., Wang, T., 2020. Failure mechanism of TRSS mode in landslides induced by earthquake. Sci. Rep. 10 (1), 21326.

[28] Liu, Z.J., Liu, W., Peng, Y.B., 2016. Random function based spectral representation of stationary and non-stationary stochastic processes. Probab. Eng. Mech. 45, 115-126.

[29] Lundberg, S.M., Lee, S.I., 2017. A unified approach to interpreting model predictions. In: Proc. 31st Annual Conference on Neural Information Processing Systems (NIPS). Long Beach, CA, 30, pp. 4768-4777.

[30] Macedo, J., Liu, C.Y., Soleimani, F., 2021. Machine-learning-based predictive models for estimating seismically-induced slope displacements. Soil Dyn. Earthq. Eng. 148, 106795.

[31] Makdisi, F.I., Seed, H.B., 1978. Simplified procedure for estimating dam and embankment earthquake-induced deformations. J. Geotech. Eng. Div. Am. Soc. Civ. Eng. 104 (7), 849-867.

[32] Meng, X.S., Liu, X.L., Wang, Y.Y., Zhang, H., Guo, X.S., 2024. Submarine landslide susceptibility assessment integrating frequency ratio with supervised machine learning approach. Appl. Ocean Res. 153, 104237.

[33] Moradi, S., Huisman, J.A., Class, H., Vereecken, H., 2018. The effect of bedrock topography on timing and location of landslide initiation using the local factor of safety concept. Water 10 (10), 1290.

[34] Mou, L., Meng, Z., Yan, R., Li, G., Xu, Y., Zhang, L., Jin, Z., 2016. How transferable are neural networks in NLP applications? In: Proc. 2016 Conference on Empirical Methods in Natural Language Processing. Austin, Texas, pp. 479-489.

[35] Mustafa, A.M., Agha, R., Ghazalat, L., Sha’ban, T., 2024. Natural disasters detection using explainable deep learning. Intell. Syst. Appl. 23, 200430.

[36] Newmark, N.M., Mortimore, N., 1965. Effects of earthquakes on dams and embankments. Geotechnique 15 (2), 139-160.

[37] Oh, B.K., Park, Y., Park, H.S., 2020. Seismic response prediction method for building structures using convolutional neural network. Struct. Control Health Monit. 27 (5), e2519.

[38] Onyelowe, K.C., Moghal, A.A.B., Ahmad, F., Rehman, A.U., Hanandeh, S., 2024. Numerical model of debris flow susceptibility using slope stability failure machine learning prediction with metaheuristic techniques trained with different algorithms. Sci. Rep. 14 (1), 19562.

[39] Pai, L.F., Wu, H.G., Yang, T., Zhong, F.F., 2021. Study on seismic coefficient calculation method of slope seismic stability analysis. Shock Vib. 2021 (1), 9986509.

[40] Pang, R., Xu, B., Zhou, Y., Song, L.F., 2021. Seismic time-history response and system reliability analysis of slopes considering uncertainty of multi-parameters and earthquake excitations. Comput. Geotech. 136, 104245.

[41] Shakoor, A., Admassu, Y., 2010. Rock Slope Design Criteria. Ohio Department of Transportation, Office of Research and Development, Ohio, USA, p. 706.

[42] Shentu, N., Yang, J.C., Li, Q., Qiu, G.H., Wang, F., 2023. Research on the landslide prediction based on the dual mutual-inductance deep displacement 3D measuring sensor. Appl. Sci. 13 (1), 213.

[43] Song, C., Chen, B., Li, Y., Li, Z.H., Du, J.T., Yu, C., Peng, J.B., Liu, H.H., Liu, Z.J., Hu, X.N., Li, X.L., Hu, Y., Crosta, G., 2025. Amplified coseismic loess failure and postseismic landslide acceleration triggered by the 2023 Jishishan, China earthquake. Eng. Geol. 352, 108074.

[44] Song, D.Q., Shi, W.P., Wang, C.W., Dong, L.H., He, X., Wu, E.E., Zhao, J.J., Lu, R.H., 2023. Numerical investigation of a local precise reinforcement method for dynamic stability of rock slope under earthquakes using continuum-discontinuum element method. Sustainability 15 (3), 2490.

[45] Song, J., Xue, Y.L., Zhang, S., Ji, J., Gao, Y.F., 2024. Directionality of earthquake-induced slope displacements from numerical analysis and sliding block approaches. Eng. Geol. 339, 107638.

[46] Surendran, R., Anitha, J., Angelopoulou, A., Kapetanios, E., Chausalet, T., Jude Hemanth, D., 2022. Transfer Learning Based Natural Scene Classification for Scene Understanding by Intelligent Machines. In: Groen D., de Mulatier C., Paszynski M., Krzhizhanovskaya V.V., Dongarra J.J., Sloot P.M.A.(Eds.), Computional, Science - ICCS 2002. Lecture Notes in Computer Science, vol. 13351. Springer, Cham, pp. 41-48.

[47] Tang, C., Zhu, J., Qi, X., Ding, J., 2011. Landslides induced by the Wenchuan earthquake and the subsequent strong rainfall event: A case study in the Beichuan area of China. Eng. Geol. 122 (1-2), 22-33.

[48] Wang, H.J., Wang, L., Zhang, L.M., 2023. Transfer learning improves landslide susceptibility assessment. Gondwana Res. 123, 238-254.

[49] Wang, J.Y., Shahani, N.M., Zheng, X.G., Jiang, H.W., Wei, X., 2025. Machine learning-based analyzing earthquake-induced slope displacement. PLoS One 20 (2), e0314977. https://doi.org/10.1371/journal.pone.0314977.

[50] Wang, M.X., Huang, D.R., Wang, G., Li, D.Q., 2020. SS-XGBoost: A machine learning framework for predicting Newmark sliding displacements of slopes. J. Geotech. Geoenviron. Eng. 146 (9), 04020074.

[51] Wang, X.Y., Nie, G.Z., Ma, M.J., 2011. Evaluation model of landslide hazards induced by the 2008 Wenchuan earthquake using strong motion data. Earthq. Sci. 24 (3), 311-319.

[52] Wang, Y., Wang, H., Cui, P., Chen, D.L., Tang, J.B., Ou, T.H., Hao, J.S., Wang, J., Zhang, G.T., Lei, Y., Wu, C.H., 2024. Disaster effects of climate change and the associated scientific challenges. Chin. Sci. Bull. 69 (2), 286-300 (in Chinese with English abstract).

[53] Whipple, K.X., Shirzaei, M., Hodges, K.V., Arrowsmith, J.R., 2016. Active shortening within the Himalayan orogenic wedge implied by the 2015 Gorkha earthquake. Nat. Geosci. 9 (9), 711-716.

[54] Wu, M., Xu, X., Han, X., Du, X.L., 2024. Seismic performance prediction of a slope-pile-anchor coupled reinforcement system using recurrent neural networks. Eng. Geol. 338, 107623.

[55] Wyllie, D.C., Mah, C., 2004. Principles of Rock Slope Design. In: Wyllie D.C., Mah C. (Eds.),Rock Slope Engineering. 4th ed. CRC Press, Boca Raton, pp. 1-21.

[56] Xiao, S.H., Zhang, L.M., He, J., Peng, M., Jiang, R.C., Lu, W.J., 2024. Hypermobility of a catastrophic earthquake-induced loess landslide. Eng. Geol. 343, 107777.

[57] Xiong, M., Huang, Y., Zhao, Q., 2018. Effect of travelling waves on stochastic seismic response and dynamic reliability of a long-span bridge on soft soil. Bull. Earthq. Eng. 16 (9), 3721-3738.

[58] Xu, C., Dai, F.C., Xu, X.W., Lee, Y.H., 2012. GIS-based support vector machine modeling of earthquake-triggered landslide susceptibility in the Jianjiang River watershed, China. Geomorphology 145, 70-80.

[59] Xu, W.H., Kang, Y.F., Chen, L.C.A., Wang, L.Q., Qin, C.B., Zhang, L.T., Liang, D., Wu, C.Z., Zhang, W.G., 2023. Dynamic assessment of slope stability based on multi-source monitoring data and ensemble learning approaches: A case study of Jiuxianping landslide. Geol. J. 58 (6), 2353-2371.

[60] Xu, X., Xing, Y., Guo, Z., Huang, Y., 2021. Stability analysis of rainfall-triggered toe-cut slopes and effectiveness evaluation of pile-anchor structures. J. Earth Sci. 32 (5), 1104-1112.

[61] Zhang, R., Yang, Y.J., Wang, T.Y., Liu, A.M.Y., Lv, J.C., He, X., Fu, Y., Zhang, B., Dai, K.R., Liu, G.X., 2024. Co-seismic landslide susceptibility mapping for the Luding earthquake area based on heterogeneous ensemble machine learning models. Int. J. Digit. Earth 17 (1), 2409337.

[62] Zhang, R.Y., Chen, Z., Chen, S., Zheng, J.W., Büyüköztürk, O., Sun, H., 2019. Deep long short-term memory networks for nonlinear structural seismic response prediction. Comput. Struct. 220, 55-68.

[63] Zhang, W., Luo, Q., Wang, T., Wang, Q., Jiang, L., Kong, D., 2022a. System reliability evaluation of long railway subgrade slopes considering discrete instability. Railw. Eng. Sci. 30 (4), 482-493.

[64] Zhang, Y., Yang, Q., 2018. An overview of multi-task learning. Natl. Sci. Rev. 5 (1), 30-43.

[65] Zhang, Y., Yang, Q., 2022. A survey on multi-task learning. IEEE Trans. Knowl. Data Eng. 34 (12), 5586-5609.

[66] Zhang, Y.L., Du, D.S., Shi, S., Li, W.W., Wang, S.G., 2021. Effects of the earthquake nonstationary characteristics on the structural dynamic response: Base on the BP neural networks modified by the genetic algorithm. Buildings 11 (2), 69.

[67] Zhang, Z.H., He, N.X., Li, D.Y., Gao, H., Gao, T.G., Zhou, C., 2022b. Federated transfer learning for disaster classification in social computing networks. J. Saf. Sci. Resil. 3 (1), 15-23.

[68] Zhao, L.Y., Huang, Y., Hu, H.Q., 2020. Stochastic seismic response of a slope based on large-scale shaking-table tests. Eng. Geol. 277, 105782.

[69] Zhao, Y., Huang, Z., Wei, Z.L., Zheng, J., Konagai, K., 2023. Assessment of earthquake-triggered landslide susceptibility considering coseismic ground deformation. Front. Earth Sci. 10, 993975.

[70] Zheng, F., Leung, Y.F., Zhu, J.B., Jiao, Y.Y., 2019. Modified predictor-corrector solution approach for efficient discontinuous deformation analysis of jointed rock masses. Int. J. Numer. Anal. Methods Geomech. 43 (2), 599-624.

[71] Zheng, S., Jiang, A.N., Feng, K.S., 2021. A reliability evaluation method for intermittent jointed rock slope based on evolutionary support vector machine. CMES-Comput. Model. Eng. Sci. 129 (1), 149-166.

[72] Zhou, C., Cao, Y., Gan, L.L., Wang, Y., Motagh, M., Roessner, S., Hu, X., Yin, K.L., 2024. A novel framework for landslide displacement prediction using MT-InSAR and machine learning techniques. Eng. Geol. 334, 107497.

[73] Zhou, C., Yin, K.L., Cao, Y., Ahmed, B., Fu, X.L., 2018. A novel method for landslide displacement prediction by integrating advanced computational intelligence algorithms. Sci. Rep. 8 (1), 7287.

[74] Zoph, B., Yuret, D., May, J., Knight, K., 2016. Transfer Learning for Low-Resource Neural Machine Translation. Association for Computational Linguistics, Austin, Texas, pp. 1568-1575.