{"id":2130,"date":"2025-11-30T07:41:23","date_gmt":"2025-11-30T07:41:23","guid":{"rendered":"https:\/\/scipapermill.com\/index.php\/2025\/11\/30\/retrieval-augmented-generation-from-efficiency-to-robustness-in-the-era-of-llms\/"},"modified":"2025-12-28T21:08:35","modified_gmt":"2025-12-28T21:08:35","slug":"retrieval-augmented-generation-from-efficiency-to-robustness-in-the-era-of-llms","status":"publish","type":"post","link":"https:\/\/scipapermill.com\/index.php\/2025\/11\/30\/retrieval-augmented-generation-from-efficiency-to-robustness-in-the-era-of-llms\/","title":{"rendered":"Retrieval-Augmented Generation: From Efficiency to Robustness in the Era of LLMs"},"content":{"rendered":"

Latest 50 papers on retrieval-augmented generation: Nov. 30, 2025<\/h3>\n

The landscape of AI, particularly in Natural Language Processing, is rapidly being reshaped by the remarkable capabilities of Large Language Models (LLMs). However, these powerful models often grapple with challenges like factual accuracy, domain specificity, and computational efficiency. This is where Retrieval-Augmented Generation (RAG) steps in, offering a dynamic solution by grounding LLM responses in external, up-to-date knowledge bases. Recent breakthroughs, as showcased in a collection of cutting-edge research, are pushing the boundaries of RAG, addressing critical aspects from efficiency and multi-modality to domain adaptation and security.<\/p>\n

The Big Idea(s) & Core Innovations<\/h3>\n

One of the central themes emerging from recent research is the drive for smarter, more efficient knowledge retrieval<\/strong>. Traditional RAG often relies on fixed top-k<\/code> document retrieval, which can be inefficient or lead to irrelevant contexts. This challenge is directly addressed by Yifan Xu et al.\u00a0from Coinbase and USC<\/em> in their paper, \u201cCluster-based Adaptive Retrieval: Dynamic Context Selection for RAG Applications<\/a>\u201d. They introduce Cluster-based Adaptive Retrieval (CAR), which dynamically adjusts the number of retrieved documents based on query complexity, significantly reducing token usage and latency while improving relevance. Similarly, FastLM\u2019s<\/em> \u201cTowards Hyper-Efficient RAG Systems in VecDBs: Distributed Parallel Multi-Resolution Vector Search<\/a>\u201d tackles efficiency in vector databases, proposing a distributed multi-resolution search framework.<\/p>\n

The push for multi-modal RAG<\/strong> is another dominant trend. Integrating information beyond text, such as images and video, is proving crucial for richer understanding. Xiaoxing You et al.\u00a0from Hangzhou Dianzi University and Harbin Institute of Technology<\/em> present MERGE, a \u201cKnowledge Completes the Vision: A Multimodal Entity-aware Retrieval-Augmented Generation Framework for News Image Captioning<\/a>\u201d, which builds an Entity-Centric Multimodal Knowledge Base (EMKB) for precise visual-entity grounding. Following this, Xiaozhe Chen et al.\u00a0from Zhejiang University and Microsoft Research<\/em> introduce AdaVideoRAG in \u201cAdaVideoRAG: Omni-Contextual Adaptive Retrieval-Augmented Efficient Long Video Understanding<\/a>\u201d, a framework that adaptively routes retrieval strategies based on query difficulty for long video comprehension. Further, Yongdong Luo et al.\u00a0from Xiamen University and Nanjing University<\/em> in \u201cVideo-RAG: Visually-aligned Retrieval-Augmented Long Video Comprehension<\/a>\u201d achieve proprietary-level performance with open-source models for long video understanding by integrating OCR, ASR, and object detection. For visually-rich documents, Anyang Tong et al.\u00a0from Hefei University of Technology and KU Leuven<\/em> propose \u201cHKRAG: Holistic Knowledge Retrieval-Augmented Generation Over Visually-Rich Documents<\/a>\u201d, a framework that retrieves both salient and fine-print knowledge, proving essential for accurate document understanding.<\/p>\n

Specialized RAG applications<\/strong> are also gaining traction across diverse domains. In healthcare, Zhe Li et al.\u00a0from Peking Union Medical College Hospital<\/em> introduce KRAL, a \u201cKnowledge and Reasoning Augmented Learning for LLM-assisted Clinical Antimicrobial Therapy<\/a>\u201d paradigm that significantly improves diagnostic capabilities. Anonymized Author et al.\u00a0from Respiratory Medicine<\/em> also tackle medical diagnosis with \u201cLarge Language Model Aided Birt-Hogg-Dube Syndrome Diagnosis with Multimodal Retrieval-Augmented Generation<\/a>\u201d, using clinical data to reduce hallucinations. For engineering, Bingkun Guo et al.\u00a0from Zhejiang University<\/em> present an \u201cA Multidisciplinary Design and Optimization (MDO) Agent Driven by Large Language Models<\/a>\u201d, semi-automating mechanical design from natural language. RAG is even making waves in software engineering with Zhijie Chen et al.\u00a0from Nantong University<\/em> proposing \u201cReVul-CoT: Towards Effective Software Vulnerability Assessment with Retrieval-Augmented Generation and Chain-of-Thought Prompting<\/a>\u201d for enhanced software vulnerability assessment. The concept of using RAG for dynamic context generation extends to enhancing LLM efficiency, as shown by Zhan Su et al.\u00a0from Universit\u00e9 de Montr\u00e9al<\/em> in \u201cParametric Retrieval-Augmented Generation using Latent Routing of LoRA Adapters<\/a>\u201d with Poly-PRAG, which encodes documents into compact LoRA adapters for efficient retrieval.<\/p>\n

Finally, ensuring robustness and security<\/strong> in RAG systems is paramount. Badrinath Ramakrishnan and Akshaya Balaji<\/em> propose \u201cSecuring AI Agents Against Prompt Injection Attacks<\/a>\u201d, reducing attack success rates significantly. Furthermore, Yingjia Shang et al.\u00a0from Westlake University<\/em> in \u201cMedusa: Cross-Modal Transferable Adversarial Attacks on Multimodal Medical Retrieval-Augmented Generation<\/a>\u201d expose critical vulnerabilities in medical RAG systems, while Linyin Luo et al.\u00a0from The Hong Kong Polytechnic University<\/em> unveil \u201cHV-Attack: Hierarchical Visual Attack for Multimodal Retrieval Augmented Generation<\/a>\u201d, demonstrating how imperceptible visual perturbations can disrupt multimodal RAG.<\/p>\n

Under the Hood: Models, Datasets, & Benchmarks<\/h3>\n

Innovations in RAG are often powered by novel architectures, custom datasets, and rigorous benchmarks. Here\u2019s a look at some key resources:<\/p>\n