提出抗干扰量子联邦学习，仅用20%-50%客户端训练即提升15个百分点鲁棒性。
RobQFL: Robust Quantum Federated Learning in Adversarial Environment
- 将对抗训练嵌入联邦流程，通过调参控制客户端覆盖与扰动策略
- 在15个客户端上，仅20-50%对抗训练可使ε≤0.1时准确率提升约15个百分点
- 适合关注量子联邦学习安全性的研究人员和实际部署者
量子联邦学习（QFL）结合了隐私保护的联邦机制与量子计算优势，但其对对抗噪声的鲁棒性尚不明确。我们首次证明QFL与集中式量子学习一样脆弱。为此提出鲁棒量子联邦学习（RobQFL），将对抗训练直接融入联邦循环。RobQFL引入可调参数：客户端覆盖度γ（0-100%）、扰动调度（固定ε与ε混合）、优化方式（微调与从头训练），并将其生成的γ×ε空间提炼为两个指标：准确率-鲁棒性面积与鲁棒性体积。在15客户端、MNIST与Fashion-MNIST数据集上，独立同分布（IID）与非独立同分布（Non-IID）条件下，仅对20%-50%客户端进行对抗训练，即可在ε≤0.1时实现约15个百分点的准确率提升，且干净数据准确率损失低于2个百分点；微调可再提升3-5个百分点。当客户端覆盖≥75%时，中等ε混合策略最优；高ε调度仅在100%覆盖下有效。按标签排序的非独立同分布数据使鲁棒性降低一半，凸显数据异质性是主要风险。
原文摘要 · Abstract (English)
Quantum Federated Learning (QFL) merges privacy-preserving federation with quantum computing gains, yet its resilience to adversarial noise is unknown. We first show that QFL is as fragile as centralized quantum learning. We propose Robust Quantum Federated Learning (RobQFL), embedding adversarial training directly into the federated loop. RobQFL exposes tunable axes: client coverage $γ$ (0-100\%), perturbation scheduling (fixed-$\varepsilon$ vs $\varepsilon$-mixes), and optimization (fine-tune vs scratch), and distils the resulting $γ\times \varepsilon$ surface into two metrics: Accuracy-Robustness Area and Robustness Volume. On 15-client simulations with MNIST and Fashion-MNIST, IID and Non-IID conditions, training only 20-50\% clients adversarially boosts $\varepsilon \leq 0.1$ accuracy $\sim$15 pp at $< 2$ pp clean-accuracy cost; fine-tuning adds 3-5 pp. With $\geq$75\% coverage, a moderate $\varepsilon$-mix is optimal, while high-$\varepsilon$ schedules help only at 100\% coverage. Label-sorted non-IID splits halve robustness, underscoring data heterogeneity as a dominant risk.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。