量子核学习中,精准分配测量次数提升分类精度。
AQKA: Active Quantum Kernel Acquisition Under a Shot Budget

- 按重要性动态分配测量资源,而非均匀分配。
- 在225到1000个样本下,精度比均匀分配高8至26个百分点。
- 适用于量子硬件上实时自适应的高效训练,适合资源受限场景。
在近中期量子硬件上估计一个 $N \times N$ 的量子核需 $Θ(N^2 S)$ 次测量,是部署的主要瓶颈。现有方法(如 Nyström-QKE、ShoFaR、核目标对齐)仅选择测量哪些条目,但对选定子集内采用均匀分配测量次数,忽略了每个条目对下游分类器的贡献度。本文提出两种贡献:第一,构建了测量预算下的完整策略分解框架,明确了各分配策略的适用场景;所提方法 AQKA 在预算有限($B \lesssim 16 n_{\mathrm{pairs}}$)时,于稀疏敏感型KRR任务中显著优于均匀分配,精度提升从+8增至+25个百分点,且在156量子比特的 ibm_pittsburgh 硬件核上达+26至+32点;当预算充足时,Nyström-QKE 在植根稀疏任务中凭借低秩重构占优;ShoFaR 仅在极端低预算下具竞争力。第二,提出闭式配对级采集理论:最优测量数 $s_{ij}^{\star} \propto |g_{ij}|\sqrt{K_{ij}(1-K_{ij})}$,其中 $g_{ij}$ 为显式梯度(如KRR中 $|β_iα_j+β_jα_i|\sqrt{K_{ij}(1-K_{ij})}$,SVM中通过包络定理得 $|η_i^*η_j^*|\sqrt{K_{ij}(1-K_{ij})}$);修正了稀疏感知的Cauchy–Schwarz速率 $ρ \le 2m/N$,匹配实证结果(非朴素 $m^2/N^2$);给出可插值的显式常数后悔界(定理2);并推导更紧的SVM上界 $ρ^{\mathrm{SVM}} \le m_{\mathrm{sv}}^2/N^2$。最后,在真实量子硬件上首次实现多种子在线自适应测量分配,于 $N=20$ 的 ibm_aachen 上获得 $+17.0 \pm 4.8$ 点增益(5种子,$3.5σ$),且在 $N=30$ 的 ibm_berlin 上高预算条件下仍保持 $+14.0 \pm 8.5$ 点优势。
原文摘要 · Abstract (English)
Estimating an $N \times N$ quantum kernel from circuit fidelities requires $Θ(N^2 S)$ measurement shots, the dominant bottleneck for deployment on near-term hardware. Existing budget-saving methods (Nyström-QKE, ShoFaR, kernel-target alignment) sub-sample \emph{which} entries to measure but allocate shots \emph{uniformly} within their chosen subset, ignoring how much each entry drives the downstream classifier. We close this gap with two contributions. \textbf{First, a complete regime decomposition} for shot-budgeted quantum kernel learning: a principled menu of when each allocator wins. Our method, \emph{AQKA}, dominates the budget-limited regime ($B \lesssim 16 n_{\mathrm{pairs}}$) on sparse-sensitivity KRR, with the gap \emph{growing} from $+8$ to $+25$ pts over uniform as $N$ scales $225{\to}1000$ and reaching $+26$--$32$ pts on an \texttt{ibm\_pittsburgh} (156-qubit Heron) hardware kernel; Nyström-QKE wins at saturating budgets on planted-sparse via low-rank reconstruction; ShoFaR is competitive only at extreme low budgets. \textbf{Second, a closed-form pair-level acquisition theory}: $s_{ij}^{\star} \propto |g_{ij}|\sqrt{K_{ij}(1-K_{ij})}$ with explicit gradient $g_{ij}$ for KRR (Lemma~1, $|β_iα_j+β_jα_i|\sqrt{K_{ij}(1-K_{ij})}$) and SVM via the envelope theorem ($|η_i^*η_j^*|\sqrt{K_{ij}(1-K_{ij})}$); a \emph{corrected} sparsity-aware Cauchy--Schwarz rate $ρ\le 2m/N$ matching empirics (vs.\ the naive $m^2/N^2$); an explicit-constant plug-in regret bound (Theorem~2); and a tighter SVM ceiling $ρ^{\mathrm{SVM}} \le m_{\mathrm{sv}}^2/N^2$. We close with the first multi-seed live online adaptive shot allocation on quantum hardware: $+17.0 \pm 4.8$ pts at $N{=}20$ on \texttt{ibm\_aachen} ($3.5σ$, 5 seeds), with the advantage holding at $N{=}30$ at higher budget on \texttt{ibm\_berlin} ($+14.0 \pm 8.5$ pts, 5 seeds).
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。