DyLoC: 一种双层架构用于在多项式-DLA约束下的安全且可训练的量子机器学习
DyLoC: A Dual-Layer Architecture for Secure and Trainable Quantum Machine Learning Under Polynomial-DLA constraint
AI总结:
DyLoC通过双层架构在多项式-DLA约束下实现安全且可训练的量子机器学习,利用正交解耦策略提升隐私保护与训练效率。
摘要:
变分量子电路面临隐私与可训练性之间的关键权衡。为获得稳健隐私所需的高表达性会诱导指数级大的动态李代数结构,这种结构不可避免地导致 barren plateaus。相反,受限制于多项式大小的代数结构的可训练模型对代数攻击是透明的。为解决这一困境,提出了DyLoC。这种双层架构采用正交解耦策略。可训练性锚定于多项式-DLA假设,而隐私则外部化到输入和输出接口。具体而言,截断切比雪夫图编码(TCGE)用于防止快照逆向。动态局部散乱(DLS)用于混淆梯度。实验表明,DyLoC在基线水平上收敛,最终损失为0.186。它通过将梯度重建误差增加13个数量级来超越基线。此外,当重建均方误差超过2.0时,快照逆向攻击被阻止。这些结果证实了DyLoC有效建立了安全且可训练的量子机器学习的可验证路径。
英文摘要:
Variational quantum circuits face a critical trade-off between privacy and trainability. High expressivity required for robust privacy induces exponentially large dynamical Lie algebras. This structure inevitably leads to barren plateaus. Conversely, trainable models restricted to polynomial-sized algebras remain transparent to algebraic attacks. To resolve this impasse, DyLoC is proposed. This dual-layer architecture employs an orthogonal decoupling strategy. Trainability is anchored to a polynomial-DLA ansatz while privacy is externalized to the input and output interfaces. Specifically, Truncated Chebyshev Graph Encoding (TCGE) is employed to thwart snapshot inversion. Dynamic Local Scrambling (DLS) is utilized to obfuscate gradients. Experiments demonstrate that DyLoC maintains baseline-level convergence with a final loss of 0.186. It outperforms the baseline by increasing the gradient reconstruction error by 13 orders of magnitude. Furthermore, snapshot inversion attacks are blocked when the reconstruction mean squared error exceeds 2.0. These results confirm that DyLoC effectively establishes a verifiable pathway for secure and trainable quantum machine learning.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。