Abstract Currently, data privacy and security are critical in deep learning, especially with biometric data. Federated learning (FL) addresses these concerns, but struggles with security, complexity, and high-dimensional data. Quantum Federated Learning (QFL) overcomes these issues by combining FL with quantum computing, thereby offering greater power and parallelism. In this paper, we propose an innovative QFL approach with a quantum–classical hybrid architecture to address the limitations of classical FL and achieve superior performance in video and image classification. The proposed architecture consists of a classical server and quantum–classical clients. On the client side, after preprocessing the videos and images, a pre-trained model extracts high-level features, which are then classified by a client-specific quantum circuit using a Variational Quantum Circuit (VQC). To maximize the performance of this hybrid system, we jointly optimize the VQC parameters and the weights of the pre-trained model’s final feature-extraction layers. The proposed approach has been extensively evaluated in a simulation environment with random client selection to account for the heterogeneity of real-world systems. Furthermore, we present results obtained from quantum computers using globally optimized VQC parameters. This successful validation on actual quantum hardware underscores the practical applicability of our approach, extending its significance beyond theoretical contributions to next-generation applications such as smart campus security.
Abstract Currently, data privacy and security are critical in deep learning, especially with biometric data. Federated learning (FL) addresses these concerns, but struggles with security, complexity, and high-dimensional data. Quantum Federated Learning (QFL) overcomes these issues by combining FL with quantum computing, thereby offering greater power and parallelism. In this paper, we propose an innovative QFL approach with a quantum–classical hybrid architecture to address the limitations of classical FL and achieve superior performance in video and image classification. The proposed architecture consists of a classical server and quantum–classical clients. On the client side, after preprocessing the videos and images, a pre-trained model extracts high-level features, which are then classified by a client-specific quantum circuit using a Variational Quantum Circuit (VQC). To maximize the performance of this hybrid system, we jointly optimize the VQC parameters and the weights of the pre-trained model’s final feature-extraction layers. The proposed approach has been extensively evaluated in a simulation environment with random client selection to account for the heterogeneity of real-world systems. Furthermore, we present results obtained from quantum computers using globally optimized VQC parameters. This successful validation on actual quantum hardware underscores the practical applicability of our approach, extending its significance beyond theoretical contributions to next-generation applications such as smart campus security.
In recent years, the growing demand for secure, efficient, and scalable machine learning has highlighted significant challenges in traditional Federated Learning (FL) systems. These challenges include computational inefficiencies in handling large-scale, high-dimensional data, privacy concerns, and the limitations of classical hardware in solving complex optimization problems. Additionally, the increasing complexity of data and the need for more robust privacy-preserving methods have pushed the boundaries of classical FL. To address these issues, this paper presents a detailed framework for Quantum Federated Learning (QFL), where quantum computing and FL converge to enable distributed quantum model training with enhanced privacy preservation. The system is modeled using local quantum models on clients and a global aggregation strategy at the server. We develop quantum gradient-based optimization with quantum cross-entropy loss, privacy preservation through encryption, and robustness to quantum noise. A set of theorems is provided to prove the correctness and efficiency of the model, supported by rigorous mathematical formulations. We used IBM Qiskit for the simulation and compared our proposed QFL with classical FL. The results demonstrate that our QFL outperformed classical FL in terms of accuracy, precision, recall, and F1 score.
Federated learning (FL) has emerged as a promising technique to realize distributed machine learning (ML) in practice. FL enables multiple clients to collaboratively train a common ML model without the need to collect raw data from clients, which therefore has merit in the protection of data privacy. On the other hand, it is known that quantum computing excels in solving specific problems that are computationally prohibitive on classical computers due to its ability to harness quantum superposition and entanglement. However, current noisy intermediate-scale quantum (NISQ) computers have difficulties in dealing with many-qubit computation due to the lack of reliable error-correction schemes, which renders hybrid quantum-classical computing with few qubits a promising alternative. In this paper, we investigate how hybrid quantum-classical computing can be applied to FL while considering the data heterogeneity among clients. To this end, we propose the two-qubit quantum circuit-embedded convolutional neural network (2QCNN) for each client, which incorporates parallel two-qubit variational quantum circuits (VQCs) between the fully connected layers of the CNN. Simulation results show that 2QCNN outperforms the comparison schemes in terms of test accuracies. In addition, the impacts of the number and depth of parallel two-qubit VQCs on the performance of 2QCNN under various degrees of data heterogeneity are further evaluated.
Anomaly detection has a significant impact on applications such as video surveillance, medical diagnostics, and industrial monitoring, where anomalies frequently depend on context and anomaly-labeled data are limited. Quantum federated learning (QFL) overcomes these concerns by distributing model training among several quantum clients, consequently eliminating the requirement for centralized quantum storage and processing. However, in real-life quantum networks, clients frequently differ in terms of hardware capabilities, circuit designs, noise levels, and how classical data is encoded or preprocessed into quantum states. These differences create inherent heterogeneity across clients - not just in their data distributions, but also in their quantum processing behaviors. As a result, training a single global model becomes ineffective, especially when clients handle imbalanced or non-identically distributed (non-IID) data. To address this, we propose a new framework called personalized quantum federated learning (PQFL) for anomaly detection. PQFL enhances local model training at quantum clients using parameterized quantum circuits and classical optimizers, while introducing a quantum-centric personalization strategy that adapts each client's model to its own hardware characteristics and data representation. Extensive experiments show that PQFL significantly improves anomaly detection accuracy under diverse and realistic conditions. Compared to state-of-the-art methods, PQFL reduces false errors by up to 23%, and achieves gains of 24.2% in AUROC and 20.5% in AUPR, highlighting its effectiveness and scalability in practical quantum federated settings.
The exponential growth of bioinformatics data, including genomic sequences and protein structures, necessitates advanced analytical frameworks that ensure privacy, computational efficiency, and scalability. To address these challenges, we propose an integrated framework that combines Federated Learning (FL) with Quantum Computing (QC) for secure and efficient bioinformatics data analysis across distributed healthcare institutions. The proposed Quantum Federated Learning (QFL) architecture employs variational quantum circuits to expedite local model training, achieving a 35% reduction in computation time compared to classical FL methods. To strengthen security, Quantum Secure Communication Protocols (QSCP) are incorporated, resulting in over 40% enhanced resilience against adversarial attacks relative to conventional encryption schemes. Experimental validation on distributed genomic datasets, such as the Cancer Genome Atlas, demonstrates that QFL attains a classification accuracy of 94.8% in gene expression-based cancer subtype prediction, surpassing traditional FL by 6.3%. Furthermore, the model exhibits a 25% reduction in communication overhead through quantum state compression, ensuring efficient data exchange. This hybrid approach effectively addresses the critical requirements of data privacy, computational acceleration, and model robustness in bioinformatics, establishing a promising pathway for secure and scalable precision medicine solutions empowered by quantum-enhanced federated learning.
Federated Learning (FL) has emerged as a promising approach for collaborative model training across decentralized devices while addressing privacy concerns and the impracticality of centralized learning. At the same time, Quantum Machine Learning (QML) has gained considerable attention for its potential advantages in various tasks. This intersection has given rise to the field known as Quantum Federated Learning (QFL). However, noisy quantum hardware and the computational burden of datasets containing large feature spaces still pose significant challenges. To address these issues, we present what we believe to be the first approach to Vertical Quantum Federated Learning (VQFL), a QFL scheme where features, rather than data samples, are distributed among multiple parties. Our approach is capable of utilizing fully quantum models without relying on any classical layers or classical data reduction techniques. Moreover, its scalability makes it a promising candidate for leveraging current Noise Intermediate-Scale Quantum (NISQ) devices. We successfully perform 10-class classification while preserving privacy, simulating a realistic scenario where data features are randomly distributed across parties. Our experiments can be found in https://github.com/roccobb/VQFL.
Quantum federated learning (QFL) combines quantum computing and federated learning to enable decentralized model training while maintaining data privacy. QFL can improve computational efficiency and scalability by taking advantage of quantum properties such as superposition and entanglement. However, existing QFL frameworks largely focus on homogeneity among quantum clients, and they do not account for real-world variances in quantum data distributions, encoding techniques, hardware noise levels, and computational capacity. These differences can create instability during training, slow convergence, and reduce overall model performance. In this paper, we conduct an in-depth examination of heterogeneity in QFL, classifying it into two categories: data or system heterogeneity. Then we investigate the influence of heterogeneity on training convergence and model aggregation. We critically evaluate existing mitigation solutions, highlight their limitations, and give a case study that demonstrates the viability of tackling quantum heterogeneity. Finally, we discuss potential future research areas for constructing robust and scalable heterogeneous QFL frameworks.