A Hybrid Quantum Neural Network for Split Learning

2024-09-25

arXiv.org

AI Summary

To address the dual challenges of resource-constrained clients incapable of executing quantum computations and the vulnerability of classical split learning to server-side data reconstruction attacks, this paper proposes Hybrid Quantum Split Learning (HQSL). HQSL enables collaborative training between classical clients and a hybrid quantum server, introducing the first split-learning–oriented hybrid quantum architecture. It features qubit-efficient quantum state encoding, parameterized quantum circuits tailored for distributed learning, and a novel noise injection mechanism to mitigate reconstruction attacks. Evaluated on Fashion-MNIST and Speech Commands, HQSL achieves >3% higher accuracy and >1.5% improvement in F1-score over classical and quantum baselines, while scaling to hundreds of clients. Empirical results demonstrate that HQSL significantly enhances both model performance and adversarial robustness—without compromising privacy guarantees.

Technology Category

• Machine Learning: Quantum Machine Learning
• Search and Optimization: Learning to Search
• Computer Vision: Adversarial Attacks & Robustness

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Security and Privacy: Security and privacy of machine learning and AI applications
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Quantum Machine Learning (QML) is an emerging field of research with potential applications to distributed collaborative learning, such as Split Learning (SL). SL allows resource-constrained clients to collaboratively train ML models with a server, reduce their computational overhead, and enable data privacy by avoiding raw data sharing. Although QML with SL has been studied, the problem remains open in resource-constrained environments where clients lack quantum computing capabilities. Additionally, data privacy leakage between client and server in SL poses risks of reconstruction attacks on the server side. To address these issues, we propose Hybrid Quantum Split Learning (HQSL), an application of Hybrid QML in SL. HQSL enables classical clients to train models with a hybrid quantum server and curtails reconstruction attacks. In addition, we introduce a novel qubit-efficient data-loading technique for designing a quantum layer in HQSL, minimizing both the number of qubits and circuit depth. Experiments on five datasets demonstrate HQSL's feasibility and ability to enhance classification performance compared to its classical models. Notably, HQSL achieves mean improvements of over 3% in both accuracy and F1-score for the Fashion-MNIST dataset, and over 1.5% in both metrics for the Speech Commands dataset. We expand these studies to include up to 100 clients, confirming HQSL's scalability. Moreover, we introduce a noise-based defense mechanism to tackle reconstruction attacks on the server side. Overall, HQSL enables classical clients to collaboratively train their models with a hybrid quantum server, leveraging quantum advantages while improving model performance and security against data privacy leakage-related reconstruction attacks.

Problem

• Enables classical clients to train with hybrid quantum servers
• Reduces data privacy leakage in split learning environments
• Minimizes qubit and circuit depth in quantum layer design

Innovation

• Hybrid Quantum Split Learning for classical clients
• Qubit-efficient data-loading technique minimizes resources
• Noise-based defense against reconstruction attacks