Gespeichert in:
| Hauptverfasser: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | https://arxiv.org/abs/2510.08976 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866911480724586496 |
|---|---|
| author | Li, Maoliang Li, Ke Liu, Yaoyang Chen, Jiayu Zheng, Zihao Wu, Yinjun Liu, Chenchen Chen, Xiang |
| author_facet | Li, Maoliang Li, Ke Liu, Yaoyang Chen, Jiayu Zheng, Zihao Wu, Yinjun Liu, Chenchen Chen, Xiang |
| contents | To effectively leverage user-specific data, retrieval augmented generation (RAG) is employed in multimodal large language model (MLLM) applications. However, conventional retrieval approaches often suffer from limited retrieval accuracy. Recent advances in multi-vector retrieval (MVR) improve accuracy by decomposing queries and matching against segmented images. They still suffer from sub-optimal accuracy and efficiency, overlooking alignment between the query and varying image objects and redundant fine-grained image segments. In this work, we present an efficient scheduling framework for image retrieval - MIRAGE. First, we introduce a novel hierarchical paradigm, employing multiple intermediate granularities for varying image objects to enhance alignment. Second, we minimize redundancy in retrieval by leveraging cross-hierarchy similarity consistency and hierarchy sparsity to minimize unnecessary matching computation. Furthermore, we configure parameters for each dataset automatically for practicality across diverse scenarios. Our empirical study shows that, MIRAGE not only achieves substantial accuracy improvements but also reduces computation by up to 3.5 times over the existing MVR system. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_08976 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | MIRAGE: Runtime Scheduling for Multi-Vector Image Retrieval with Hierarchical Decomposition Li, Maoliang Li, Ke Liu, Yaoyang Chen, Jiayu Zheng, Zihao Wu, Yinjun Liu, Chenchen Chen, Xiang Computer Vision and Pattern Recognition Distributed, Parallel, and Cluster Computing Information Retrieval To effectively leverage user-specific data, retrieval augmented generation (RAG) is employed in multimodal large language model (MLLM) applications. However, conventional retrieval approaches often suffer from limited retrieval accuracy. Recent advances in multi-vector retrieval (MVR) improve accuracy by decomposing queries and matching against segmented images. They still suffer from sub-optimal accuracy and efficiency, overlooking alignment between the query and varying image objects and redundant fine-grained image segments. In this work, we present an efficient scheduling framework for image retrieval - MIRAGE. First, we introduce a novel hierarchical paradigm, employing multiple intermediate granularities for varying image objects to enhance alignment. Second, we minimize redundancy in retrieval by leveraging cross-hierarchy similarity consistency and hierarchy sparsity to minimize unnecessary matching computation. Furthermore, we configure parameters for each dataset automatically for practicality across diverse scenarios. Our empirical study shows that, MIRAGE not only achieves substantial accuracy improvements but also reduces computation by up to 3.5 times over the existing MVR system. |
| title | MIRAGE: Runtime Scheduling for Multi-Vector Image Retrieval with Hierarchical Decomposition |
| topic | Computer Vision and Pattern Recognition Distributed, Parallel, and Cluster Computing Information Retrieval |
| url | https://arxiv.org/abs/2510.08976 |