动态专家环境下在线学习选择性调用，提升预测可靠性。
Online Learning-to-Defer with Varying Experts
- 基于在线贝叶斯反馈构建多专家动态路由策略
- 理论证明误差随时间增长为T^{2/3}阶，优于传统方法
- 适合实时系统中专家随时变化的场景
学习-去偏（L2D）方法将每个查询选择交给预测模型或外部专家处理。真实应用需应对流式数据、专家可用性变化、可靠性波动，且仅能获得所选动作的反馈。本文提出一种在线多分类L2D算法，结合查询动作的带臂反馈与动态变化的专家池。设总专家数为N=n+n_e，线性得分矩阵的Frobenius范数受B限制，增强输入范数受ρ限制。在假设线性校准和投影比较类零代理最小化间隙的前提下，该方法实现期望真去偏遗憾为O((BN^{3/2}ρ+1)T^{2/3})，在集中得分条件下可优化至O(BN^{3/2}ρ√T + B^2N^3ρ^2)。分析融合了在线H-一致性转移界与投影在线凸优化。在合成及真实数据集上的实验验证了在专家可用性和可靠性变化下仍可实现有效选择性路由。
原文摘要 · Abstract (English)
Learning-to-Defer (L2D) methods route each query either to a predictive model or to external experts. Real-world deployments require handling streaming data, changing expert availability, shifting expert reliability, and feedback observed only for the selected action. We introduce an online multiclass L2D algorithm that combines queried-action bandit feedback with a dynamically varying pool of experts. Let $N=n+n_e$, let $B$ bound the Frobenius norm of the linear score matrix, and let $ρ$ bound the augmented input norm. Assuming linear calibration and zero surrogate minimizability gap for the projected comparator class, our method achieves expected true-deferral regret $O((BN^{3/2}ρ+1)T^{2/3})$, improving to $O(BN^{3/2}ρ\sqrt T+B^2N^3ρ^2)$ under a concentrated-score condition. The analysis combines an online $\mathcal H$-consistency transfer bound with projected online convex optimization. Experiments on synthetic and real-world datasets demonstrate selective routing under varying expert availability and reliability.
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。