CoRA-NAS：用于神经架构搜索的粗排序与锚点残差细化

中文摘要 AI 辅助

零成本代理以低成本对架构进行排序，但其可靠性在不同搜索空间上存在差异。我们提出了CoRA-NAS（粗排序+锚点残差），一个两阶段框架，将静态排序先验与低成本的 learning-curve 细化相结合。CoRA-Rank 通过等权重的 log-rank 共识和无目标共识门控，聚合初始状态下的容量和结构代理。CoRA-Refine 在该先验上采样锚点，外推其早期验证曲线，并使用 ExtraTrees 模型传播学习到的残差校正。细化过程仅需完全训练候选集成本的约1%。完全训练的架构-准确率标签不用于拟合排序器。一种配置跨空间使用，并采用空间特定的架构编码。在 NAS-Bench-201、NAS-Bench-101、TransNAS-Bench-101 和 NATS-SSS 上，CoRA-Refine 的平均 Spearman 相关系数分别达到 0.946、0.715、0.786 和 0.894。其最差空间相关系数 0.715 是所比较方法中最高的。在 NAS-Bench-201/CIFAR-100 上，其选中的架构达到 73.32% 的准确率，接近报告的真实最佳值 73.37%。在纯规模空间上，细化恢复了静态先验相对于参数数量的不足，同时与最强的容量代理在噪声范围内保持并列。所得到的框架结合了跨空间排序鲁棒性与低成本架构选择。

英文摘要

Zero-cost proxies rank architectures cheaply, but their reliability varies across search spaces. We introduce CoRA-NAS (COarse Ranking + Anchor-residual), a two-stage framework combining a static ranking prior with low-cost learning-curve refinement. CoRA-Rank aggregates capacity and structure-at-initialization proxies through an equal-weight log-rank consensus and a target-free consensus gate. CoRA-Refine samples anchors across this prior, extrapolates their early validation curves, and propagates a learned residual correction with an ExtraTrees model. The refinement uses approximately 1% of the cost of fully training the candidate set. Fully trained architecture-accuracy labels are not used to fit the ranker. One configuration is used across spaces, with space-specific architecture encodings. Across NAS-Bench-201, NAS-Bench-101, TransNAS-Bench-101, and NATS-SSS, CoRA-Refine achieves mean Spearman correlations of 0.946, 0.715, 0.786, and 0.894, respectively. Its worst-space correlation of 0.715 is the highest among the compared methods. On NAS-Bench-201/CIFAR-100, its selected architecture reaches 73.32% accuracy, near the reported ground-truth best of 73.37%. On the pure size space, refinement recovers the static prior's shortfall relative to parameter count, while remaining tied with the strongest capacity proxies within noise. The resulting framework combines cross-space ranking robustness with low-cost architecture selection.

发表机构

• University of New South Wales（新南威尔士大学）
• Korea Advanced Institute of Science and Technology（韩国科学技术院）

机构由 AI 辅助整理，请以论文原文为准。