Hybrid Quantum and Long-Term and Short-Term Memory Network for Prediction of Seismic Accelerations

Fuentes Huaman Y.; Cardenas Solano J.; Calderon Aedo K.J.; Huancahuire Bravo C.; Yanqui Daz F.; Ponce Torres E.; Aguirre Ramos F.; Aquino Cruz M.A.; Zegarra Ramirez L.E. · ISMSI 2026 - 2026 10th International Conference on Intelligent Systems, Metaheuristics and Swarm Intelligence (2026) · DOI: 10.1145/3820820.3820848

Resumen

The study evaluated the performance of a traditional Long Short-Term Memory (LSTM) network and a hybrid Quantum-LSTM model for predicting seismic accelerations in three orthogonal directions: Z (vertical), N (north-south), and E (east-west). Evaluation metrics included Mean Absolute Error (MAE), Root Mean Square Error (RMSE), Mean Square Error (MSE), coefficient of determination (R2), and Symmetric Mean Absolute Percentage Error (SMAPE). The traditional LSTM achieved RMSE values of 0.03056 (Z), 0.03737 (N), and 0.03730 (E), with R2 scores of 0.80704, 0.88248, and 0.87137, and MAE between 0.01809 and 0.02167. The Quantum-LSTM notably improved results: RMSE dropped to 0.02025 (Z), 0.01949 (N), and 0.02000 (E); R2 rose to 0.91452, 0.93287, and 0.92641; and MAE narrowed to 0.01095-0.01182.

Instituciones: Universidad Nacional Micaela Bastidas de Apurímac; Universidad Tecnológica de los Andes