Abstract: Wave-induced transient liquefaction seepage leads to the transport of fine-grained sediments from the seabed to seawater. The contribution of wave-induced transient liquefaction to the sediment resuspension cannot be ignored,and the accurate prediction of the contribution is difficult. The observation data in the Yellow River subaquatic delta include water depth,significant wave height,significant wave period,suspended sediment concentration in the benthic chamber,and suspended sediment concentration out of the benthic chamber. These data are treated as the input data sets. Then,a deep learning model of transient liquefaction contribution to resuspension is developed based on a long short-term memory recurrent neural network. To evaluate the performance of the model,the prediction results of the deep learning model based on LSTM are compared with other models. These models include Support Vector Regression and Artificial Neural Network,and have mean absolute percentage error(MAPE),root mean square error(RMSE) and mean squared error-standard deviation(RSR). The results show that the LSTM model has the smallest error for transient pumping resuspension contribution within 3.5 days,with the mean values of MAPE,RMSE,and RSR of 5.87%,1.6730,and 0.1574,respectively. Therefore,the LSTM model can effectively reduce the error superposition problem arising from machine learning methods in continuous forecasting.