Jan 2026· IEEE International Conference on Acoustics, Speech, and Signal Processing· Vol abs/2602.00104· 4 citations· 23 references
Computer Science
TL;DR
R3G, a modular Reasoning-Retrieval-Reranking framework, first produces a brief reasoning plan that specifies the required visual cues, then adopts a two-stage strategy, with coarse retrieval followed by fine-grained reranking, to select evidence images.
Abstract
Vision-centric retrieval for VQA requires retrieving images to supply missing visual cues and integrating them into the reasoning process. However, selecting the right images and integrating them effectively into the model's reasoning remains challenging. To address this challenge, we propose R3G, a modular Reasoning-Retrieval-Reranking framework. It first produces a brief reasoning plan that specifies the required visual cues, then adopts a two-stage strategy, with coarse retrieval followed by fine-grained reranking, to select evidence images. On MRAG-Bench, R3G improves accuracy across six MLLM backbones and nine sub-scenarios, achieving state-of-the-art overall performance. Ablations show that sufficiency-aware reranking and reasoning steps are complementary, helping the model both choose the right images and use them well. We release code and data at https://github.com/czh24/R3G.
This work compares six off-the-shelf VLMs against a transparent baseline that uses classical SIFT keypoint matching and a signal-quality index over retrieved images, and shows that a learned visual representation is not the bottleneck for this task: given retrieved images, a shape-appropriate classical similarity suffi...
Dynamic Multi-Path Retrieval for KB-VQA (DMRAG) is proposed, which re-trieves candidates through multiple retrieval paths that capture complementary visual and semantic cues and performs Question-Adaptive Gated Fusion to balance contributions from different modalities according to the query’s information need.
Zeyu Song, Yimin Deng, Yu-Xin Zhang et al.· Proceedings of the Thirty-Fi...· 0 citations
GEM is presented, a generative embedding model that augments retrieval through its own knowledge by explicitly reasoning about user intent and relevance criteria, and its generative nature allows test-time compute scaling via prompting to further enhance retrieval performance.
CERES is proposed, a closed-loop multimodal indexing framework that builds a three-level semantic pyramid, mines implicit concepts via a co-occurrence-aware router, performs scale-routed cross-attention into a lightweight U-Net generator, and verifies coverage by re-indexing the generated image with the same frozen VLM...
Guang-Yuan Dong, Chuang Liu, Hao-Yu Wang et al.· 3 citations
This work proposes Trident, with two complementary components: Trident-R, a retriever-agnostic LLM reranker that converts each candidate into an LLM-readable semantic record, then performs a single adaptive-K rerank call; and Trident-S, a generation-side module that prompts the VLM under topical, entity, and structural...
Guanchen Wu, Jia-Yuan Ding, Subhabrata Mukherjee et al.· 0 citations
A novel unified retrieval-augmented framework, URA-NER, including three key components: Progressive Granularity Retrieval, Model-aware Representation Enhancement, and Reason-aware Knowledge Verification is proposed, including three key components: Progressive Granularity Retrieval, Model-aware Representation Enhancemen...
Known for his clear and elegant writing style, Bertsekas shaped fields from control and optimization to large-scale computation and artificial intelligence.
