Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

arXiv:2505.09395 · IEEE International Conference on Quantum Computing and Engineering (QCE)

Quantum-Enhanced Parameter-Efficient Learning for Typhoon Trajectory Forecasting

Chen-Yu Liu, Kuan‐Cheng Chen, Yi‐Chien Chen, Samuel Yen-Chi Chen, Wei‐Hao Huang, Wei‐Jia Huang, Yen-Jui Chang

Published Aug 30, 2025• 6 citations
arXivHTMLPDF

Abstract

Typhoon trajectory forecasting is essential for disaster preparedness but remains computationally demanding due to the complexity of atmospheric dynamics and the resource requirements of deep learning models. Quantum-Train (QT), a hybrid quantum-classical framework that leverages quantum neural networks (QNNs) to generate trainable parameters exclusively during training, eliminating the need for quantum hardware at inference time. Building on QT's success across multiple domains, including image classification, reinforcement learning, flood prediction, and large language model (LLM) finetuning, we introduce Quantum Parameter Adaptation (QPA) for efficient typhoon forecasting model learning. Integrated with an Attention-based Multi-ConvGRU model, QPA enables parameterefficient training while maintaining predictive accuracy. This work represents the first application of quantum machine learning (QML) to large-scale typhoon trajectory prediction, offering a scalable and energy-efficient approach to climate modeling. Our results demonstrate that QPA significantly reduces the number of trainable parameters while preserving performance, making high-performance forecasting more accessible and sustainable through hybrid quantum-classical learning.

DOI
10.1109/qce65121.2025.00224

Related papers

Quantum-Enhanced Models for Predicting Atmospheric Dynamics in Weather Forecasting

Bakialakshmi Varadaradjaperoumal Bhuvana, Mathan S

Weather forecasting plays a critical role across various sectors, supporting strategic planning and reducing the impact of hazardous climatic conditions. However, the inherently chaotic and nonlinear nature of atmospheric systems limits the performance of conventional forecasting approaches, often resulting in prediction inaccuracies and heightened risk. This study proposes a quantum-driven predictive framework that exploits the computational strengths of Quantum Machine Learning (QML) and hybrid quantum–classical optimization to enhance the accuracy and efficiency of modern forecasting models. The framework employs Quantum Neural Networks (QNNs) combined with Variational Quantum Circuits (VQCs) to learn complex spatial–temporal dynamics embedded within large meteorological datasets. Through quantum feature encoding, key atmospheric indicators—including temperature, humidity, wind velocity, and barometric pressure—are projected into high-dimensional Hilbert spaces, enabling more expressive pattern extraction and robust predictive behavior. Simulation results reveal that the quantum-augmented approach surpasses traditional deep learning architectures in both training convergence and forecast precision when applied to extensive weather records. The hybrid design further supports scalability by intelligently distributing computational workloads between classical processors and quantum hardware. Overall, this research demonstrates the transformative potential of quantum computing in atmospheric modeling, offering a foundation for future real-time forecasting systems capable of managing the increasing complexity of global climate patterns.

Journal of Computer Applications and Information Technology•Dec 30, 2025•0 citations

Quantum-Train Long Short-Term Memory: Application on Flood Prediction Problem

Chu-Hsuan Abraham Lin, Chen-Yu Liu, Kuan‐Cheng Chen

Flood prediction is a critical challenge in the context of climate change, with significant implications for ecosystem preservation, human safety, and infrastructure protection. In this study, we tackle this problem by applying the Quantum-Train (QT) technique to a forecasting Long Short-Term Memory (LSTM) model trained by Quantum Machine Learning (QML) with significant parameter reduction. The QT technique, originally successful in the “A Matter of Taste” challenge at QHack 2024, leverages QML to reduce the number of trainable parameters to a polylogarithmic function of the number of parameters in a classical neural network (NN). This innovative framework maps classical NN weights to a Hilbert space, altering quantum state probability distributions to adjust NN parameters. Our approach directly processes classical data without the need for quantum embedding and operates independently of quantum computing resources post-training, making it highly practical and accessible for real-world flood prediction applications. This model aims to improve the efficiency of flood forecasts, ultimately contributing to better disaster preparedness and response.

arXiv:2407.08617•IEEE International Conference on Quantum Computing and Engineering (QCE)•Sep 15, 2024•39 citations

Extreme Weather Forecasting Using Quantum-Enhanced LSTM Network

Ying‐Yi Hong, Daryll John Medina, Yun-Yuan Wang

This study investigates the potential of quantum machine learning to enhance the accuracy of typhoon forecasting, with a particular emphasis on its implications for photovoltaic (PV) and wind turbine systems in the context of extreme weather preparedness. A novel approach is proposed, integrating a quantum layer into a multi-model Long Short-Term Memory (LSTM) network architecture. Implemented using CUDA-Q, the quantum layer leverages quantum phenomena to process typhoon-related data more effectively. The proposed quantum-enhanced LSTM model is evaluated on a large dataset of typhoon observations. Results demonstrate that the model achieves promising accuracy in both typhoon track forecasting—yielding mean absolute errors (MAE) of 0.0188 and 0.0307 for latitude and longitude, respectively—and intensity forecasting, with MAEs of 0.0233 for wind speed, 0.0449 for wind direction, and 0.0022 for the 50-knot wind radius. This research contributes to the growing field of quantum machine learning and its practical applications, paving the way for future advancements in typhoon prediction and disaster resilience, particularly in optimizing the operation and robustness of PV and wind turbine systems in typhoon-prone regions.

IEEE Industry Applications Society Annual Meeting (IAS)•Jun 15, 2025•3 citations

Harnessing Quantum Machine Learning to Forecast Human Movement Patterns During Natural Disasters

Huruy Bsrat, Alexander F. Wang, Hongmei Chi

Accurately forecasting how people move during natural disasters is essential for helping communities stay safe and recover quickly. Information from mobile phones, GPS, and social media gives a real time view of where people are and how they’re moving. This helps emergency teams make more time effective and accurate decisions. But traditional computer models can not handle this kind of complex and fast-changing data, especially when time is critical. This study investigates the use of quantum machine learning (QML), with a particular emphasis on quantum neural networks and hybrid quantumclassical models, to forecast mobility patterns during disasters. A structured dataset containing individual movement information within a 15 by 20 kilometer region was used to train and evaluate multiple forecasting models. The results show that the hybrid quantum-classical model achieved the highest forecasting accuracy, with a coefficient of determination ($R^{2}$) of 0.91 and a mean absolute error of approximately 1143.33 individuals. This model also maintained a good balance between predictive performance and complexity of the model, as shown in its AIC and BIC scores. In comparison, the classical multilayer perceptron performed competitive results with an $\mathbf{R}^{2}$ of $\mathbf{0. 8 9}$, while the generalized quantum model yielded an $\mathbf{R}^{2}$ of $\mathbf{0. 8 1}$. The LSTM model underperformed and it gave unstable predictions and had much higher errors. These findings highlight the potential of hybrid quantum models particularly QNN based architectures to enhance disaster preparedness by delivering accurate and scalable mobility forecasts in critical situations where early decisions are essential.

Interdisciplinary Conference on Electrics and Computer (INTCEC)•Sep 15, 2025•0 citations

Introduction to Quantum-Train Toolkit

Chen-Yu Liu, Chu-Hsuan Abraham Lin, Wei‐Jia Huang +1 more

Quantum-Train is an innovative approach that leverages quantum machine learning (QML) to train classical neural network (NN) models, offering significant parameter reduction on a polylogarithmic scale, eliminating data encoding issues, and enabling inference on purely classical computers. We introduce the Quantum-Train Toolkit, which consolidates the source code from various studies on the Quantum-Train framework. This includes image classification on MNIST, FashionMNIST, and CIFAR-10 datasets using convolutional neural networks, reinforcement learning with policy networks in Cartpole-v1 and MiniGrid environments, and Long Short-Term Memory models for flood prediction. The Quantum-Train Toolkit simplifies the execution of PyTorch multilayer perceptron models using Quantum-Train and provides guidance for adapting more complex architectures into the Quantum-Train format.

IEEE International Conference on Quantum Computing and Engineering (QCE)•Sep 15, 2024•5 citations

A Quantum Circuit-Based Compression Perspective for Parameter-Efficient Learning

Chen-Yu Liu, Chao-Han Huck Yang, Hsi‐Sheng Goan +1 more

Quantum-centric supercomputing presents a compelling framework for large-scale hybrid quantum-classical tasks. Although quantum machine learning (QML) offers theoretical benefits in various applications, challenges such as large-size data encoding in the input stage and the reliance on quantum resources in the inference stage limit its practicality for tasks like fine-tuning large language models (LLMs). Quantum parameter generation, a novel approach of QML, addresses these limitations by using quantum neural networks (QNNs) to generate classical model weights (parameters) exclusively during training, thereby decoupling inference from quantum hardware. In this work, we introduce Quantum Parameter Adaptation (QPA) in the framework of quantum parameter generation, which integrates QNNs with a classical multi-layer perceptron mapping model to generate parameters for fine-tuning methods. Using Gemma-2 and GPT-2 as case studies, QPA demonstrates significant parameter reduction for parameter-efficient fine-tuning methods, such as Low-Rank Adaptation (LoRA), while maintaining comparable or improved performance in text generation tasks. Specifically, QPA reduces the number of parameters to $52.06\%$ of the original LoRA for GPT-2 with a slight performance gain of $0.75\%$, and to $16.84\%$ for Gemma-2, with a marginal performance improvement of $0.07\%$. These results highlight QPA's ability to achieve efficient parameter reduction without sacrificing performance in the quantum parameter generation framework. This work showcases the potential of quantum-enhanced parameter reduction, offering a scalable quantum-classical solution for fine-tuning LLMs while preserving the feasibility of inference on classical hardware.

arXiv:2410.09846•Oct 13, 2024•3 citations
Typhoon trajectory forecasting is essential for disaster preparedness but remains computationally demanding due to the complexity of atmospheric dynamics and the resource requirements of deep learning models. Quantum-Train (QT), a hybrid quantum-classical framework that leverages quantum neural networks (QNNs) to generate trainable parameters exclusively during training, eliminating the need for quantum hardware at inference time. Building on QT's success across multiple domains, including image classification, reinforcement learning, flood prediction, and large language model (LLM) finetuning, we introduce Quantum Parameter Adaptation (QPA) for efficient typhoon forecasting model learning. Integrated with an Attention-based Multi-ConvGRU model, QPA enables parameterefficient training while maintaining predictive accuracy. This work represents the first application of quantum machine learning (QML) to large-scale typhoon trajectory prediction, offering a scalable and energy-efficient approach to climate modeling. Our results demonstrate that QPA significantly reduces the number of trainable parameters while preserving performance, making high-performance forecasting more accessible and sustainable through hybrid quantum-classical learning.
Weather forecasting plays a critical role across various sectors, supporting strategic planning and reducing the impact of hazardous climatic conditions. However, the inherently chaotic and nonlinear nature of atmospheric systems limits the performance of conventional forecasting approaches, often resulting in prediction inaccuracies and heightened risk. This study proposes a quantum-driven predictive framework that exploits the computational strengths of Quantum Machine Learning (QML) and hybrid quantum–classical optimization to enhance the accuracy and efficiency of modern forecasting models. The framework employs Quantum Neural Networks (QNNs) combined with Variational Quantum Circuits (VQCs) to learn complex spatial–temporal dynamics embedded within large meteorological datasets. Through quantum feature encoding, key atmospheric indicators—including temperature, humidity, wind velocity, and barometric pressure—are projected into high-dimensional Hilbert spaces, enabling more expressive pattern extraction and robust predictive behavior. Simulation results reveal that the quantum-augmented approach surpasses traditional deep learning architectures in both training convergence and forecast precision when applied to extensive weather records. The hybrid design further supports scalability by intelligently distributing computational workloads between classical processors and quantum hardware. Overall, this research demonstrates the transformative potential of quantum computing in atmospheric modeling, offering a foundation for future real-time forecasting systems capable of managing the increasing complexity of global climate patterns.
Flood prediction is a critical challenge in the context of climate change, with significant implications for ecosystem preservation, human safety, and infrastructure protection. In this study, we tackle this problem by applying the Quantum-Train (QT) technique to a forecasting Long Short-Term Memory (LSTM) model trained by Quantum Machine Learning (QML) with significant parameter reduction. The QT technique, originally successful in the “A Matter of Taste” challenge at QHack 2024, leverages QML to reduce the number of trainable parameters to a polylogarithmic function of the number of parameters in a classical neural network (NN). This innovative framework maps classical NN weights to a Hilbert space, altering quantum state probability distributions to adjust NN parameters. Our approach directly processes classical data without the need for quantum embedding and operates independently of quantum computing resources post-training, making it highly practical and accessible for real-world flood prediction applications. This model aims to improve the efficiency of flood forecasts, ultimately contributing to better disaster preparedness and response.
This study investigates the potential of quantum machine learning to enhance the accuracy of typhoon forecasting, with a particular emphasis on its implications for photovoltaic (PV) and wind turbine systems in the context of extreme weather preparedness. A novel approach is proposed, integrating a quantum layer into a multi-model Long Short-Term Memory (LSTM) network architecture. Implemented using CUDA-Q, the quantum layer leverages quantum phenomena to process typhoon-related data more effectively. The proposed quantum-enhanced LSTM model is evaluated on a large dataset of typhoon observations. Results demonstrate that the model achieves promising accuracy in both typhoon track forecasting—yielding mean absolute errors (MAE) of 0.0188 and 0.0307 for latitude and longitude, respectively—and intensity forecasting, with MAEs of 0.0233 for wind speed, 0.0449 for wind direction, and 0.0022 for the 50-knot wind radius. This research contributes to the growing field of quantum machine learning and its practical applications, paving the way for future advancements in typhoon prediction and disaster resilience, particularly in optimizing the operation and robustness of PV and wind turbine systems in typhoon-prone regions.
Accurately forecasting how people move during natural disasters is essential for helping communities stay safe and recover quickly. Information from mobile phones, GPS, and social media gives a real time view of where people are and how they’re moving. This helps emergency teams make more time effective and accurate decisions. But traditional computer models can not handle this kind of complex and fast-changing data, especially when time is critical. This study investigates the use of quantum machine learning (QML), with a particular emphasis on quantum neural networks and hybrid quantumclassical models, to forecast mobility patterns during disasters. A structured dataset containing individual movement information within a 15 by 20 kilometer region was used to train and evaluate multiple forecasting models. The results show that the hybrid quantum-classical model achieved the highest forecasting accuracy, with a coefficient of determination (R2R^{2}R2) of 0.91 and a mean absolute error of approximately 1143.33 individuals. This model also maintained a good balance between predictive performance and complexity of the model, as shown in its AIC and BIC scores. In comparison, the classical multilayer perceptron performed competitive results with an R2\mathbf{R}^{2}R2 of 0.89\mathbf{0. 8 9}0.89, while the generalized quantum model yielded an R2\mathbf{R}^{2}R2 of 0.81\mathbf{0. 8 1}0.81. The LSTM model underperformed and it gave unstable predictions and had much higher errors. These findings highlight the potential of hybrid quantum models particularly QNN based architectures to enhance disaster preparedness by delivering accurate and scalable mobility forecasts in critical situations where early decisions are essential.
Quantum-Train is an innovative approach that leverages quantum machine learning (QML) to train classical neural network (NN) models, offering significant parameter reduction on a polylogarithmic scale, eliminating data encoding issues, and enabling inference on purely classical computers. We introduce the Quantum-Train Toolkit, which consolidates the source code from various studies on the Quantum-Train framework. This includes image classification on MNIST, FashionMNIST, and CIFAR-10 datasets using convolutional neural networks, reinforcement learning with policy networks in Cartpole-v1 and MiniGrid environments, and Long Short-Term Memory models for flood prediction. The Quantum-Train Toolkit simplifies the execution of PyTorch multilayer perceptron models using Quantum-Train and provides guidance for adapting more complex architectures into the Quantum-Train format.
Quantum-centric supercomputing presents a compelling framework for large-scale hybrid quantum-classical tasks. Although quantum machine learning (QML) offers theoretical benefits in various applications, challenges such as large-size data encoding in the input stage and the reliance on quantum resources in the inference stage limit its practicality for tasks like fine-tuning large language models (LLMs). Quantum parameter generation, a novel approach of QML, addresses these limitations by using quantum neural networks (QNNs) to generate classical model weights (parameters) exclusively during training, thereby decoupling inference from quantum hardware. In this work, we introduce Quantum Parameter Adaptation (QPA) in the framework of quantum parameter generation, which integrates QNNs with a classical multi-layer perceptron mapping model to generate parameters for fine-tuning methods. Using Gemma-2 and GPT-2 as case studies, QPA demonstrates significant parameter reduction for parameter-efficient fine-tuning methods, such as Low-Rank Adaptation (LoRA), while maintaining comparable or improved performance in text generation tasks. Specifically, QPA reduces the number of parameters to 52.06%52.06\%52.06% of the original LoRA for GPT-2 with a slight performance gain of 0.75%0.75\%0.75%, and to 16.84%16.84\%16.84% for Gemma-2, with a marginal performance improvement of 0.07%0.07\%0.07%. These results highlight QPA's ability to achieve efficient parameter reduction without sacrificing performance in the quantum parameter generation framework. This work showcases the potential of quantum-enhanced parameter reduction, offering a scalable quantum-classical solution for fine-tuning LLMs while preserving the feasibility of inference on classical hardware.