相关性引导残差校准用于稠密检索嵌入谱压缩
原标题：Relevance-Guided Residual Calibration for Spectral Compression of Dense Retrieval Embeddings
AI 摘要
论文提出 Relevance-Guided Residual Calibration (RRC)，一种利用相关性标注对稠密检索嵌入的谱压缩权重进行闭式校准的方法。RRC 从正负样本得分贡献的均值与方差估计各保留方向的边际权重，将对数权重分解为幂律趋势与残差，并把残差叠加到压缩器原有对数权重上，从而在保持整体谱趋势的同时按相关性调整单个方向。实验在三个编码器、九个数据集上显示，SpecTemp+RRC 在 BEIR 与 BRIGHT 各目标维度上平均效果最佳，在 BRIGHT 五个数据集 768 维下较 SpecTemp 平均提升 2.34 nDCG@10，且在标量与乘积量化下收益仍存。
正文节选
Relevance-Guided Residual Calibration for Spectral Compression of Dense Retrieval Embeddings Abstract. Spectral compression reduces the storage and scoring costs of dense retrieval by projecting embeddings onto a smaller principal subspace. However, its corpus-derived weighting rules tie all retained directions to a shared spectral trend, without directly accounting for their individual contributions to relevance. To address this, we propose Relevance-Guided Residual Calibration (RRC), a closed-