Quantum Kernel-Based Long Short-Term Memory for Climate Time-Series Forecasting

This paper presents an enhanced QK-LSTM (Quantum Kernel-Based Long Short-Term Memory) architecture for climate time-series forecasting, particularly focused on Air Quality Index (AQI) prediction. Key contributions include:

• Embedding inputs into quantum feature spaces to capture complex nonlinear and temporal patterns more efficiently.
• Achieving fewer trainable parameters and faster convergence compared to classical and variational quantum circuit-based models.
• Scalability for deployment on NISQ-era hybrid systems, addressing environmental monitoring challenges under computational constraints.

Empirical results confirm that QK-LSTM outperforms classical LSTM models in AQI prediction, offering a practical quantum-classical approach for large-scale, high-dimensional climate data modeling.

Recommended citation: Hsu, Yu-Chao; Chen, Nan-Yow; Li, Tai-Yu; Lee, Po-Heng Henry; & Chen, Kuan-Cheng. (2025). "Quantum Kernel-Based Long Short-Term Memory for Climate Time-Series Forecasting." Proceedings of the 2025 International Conference on Quantum Communications, Networking, and Computing (QCNC), 421–426.
Download Paper