测试时自适应与查询相关残差用于视觉文档检索
Test-Time Adaptation with Query-Dependent Residuals for Visual Document Retrieval
- South China University of Technology（华南理工大学）
- Foshan University（佛山大学）
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
针对视觉文档检索中部署后难以自适应的问题，提出Q-REACT方法，利用查询相关残差将有限重排序反馈转化为可复用检索改进，在多个任务和骨干网络上提升平均检索性能且开销小。
AI中文摘要：
视觉文档检索（VDR）系统依赖于部署前计算的页面嵌入，这使得在编码器参数或语料库重新编码不可用时难以进行自适应。重排序器提供了有用的相关性信号，但传统的重排序仅将其应用于选定的查询和候选页面。我们引入了Q-REACT，一种查询侧测试时自适应方法，将有限的重排序器反馈转化为可复用的检索改进。Q-REACT学习一个共享的低秩变换，生成查询相关残差，将自适应查询得分与文档级上下文相结合，并通过在完整任务特定页面索引上归一化的学生分布来蒸馏重排序器偏好。这种设计允许未评分的页面通过缓存嵌入参与竞争，同时保持编码器和页面索引固定。在八个ViDoRe V3任务和五个开放权重及专有骨干网络上，Q-REACT在稀疏和全覆盖预算下相对于评估基线提高了平均检索性能，可迁移到保留的查询和任务，并增加了少量推理开销。结果表明，有限的重排序器反馈可以在查询集合中摊销，而无需重新训练或重建检索器。
英文摘要：
Visual document retrieval (VDR) systems depend on page embeddings computed before deployment, which makes adaptation difficult when encoder parameters or corpus re-encoding are unavailable. Rerankers provide useful relevance signals, but conventional reranking applies them only to selected queries and candidate pages. We introduce Q-REACT, a query-side test-time adaptation method that converts limited reranker feedback into reusable retrieval improvements. Q-REACT learns a shared low-rank transformation that produces query-dependent residuals, combines adapted query scores with document-level context, and distills reranker preferences with a student distribution normalized over the complete task-specific page index. This design lets unscored pages compete through cached embeddings while keeping the encoders and page index fixed. Across eight ViDoRe V3 tasks and five open-weight and proprietary backbones, Q-REACT improves average retrieval over evaluated baselines at sparse and full-coverage budgets, transfers to held-out queries and tasks, and adds little inference overhead. The results show that finite reranker feedback can be amortized across a query collection without retraining or rebuilding the retriever.
