EUVIMEDEUVIMED entdecken
Crossref

VaccineRAG: Boosting Multimodal Large Language Models’ Immunity to Harmful RAG Samples

Qixin Sun, Ziqin Wang, Hengyuan Zhao, Yilin Li, Kaiyou Song, Si Liu, Xiaolin Hu, Qingpei Guo, Linjiang Huang

Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (11) · 9189-9197

DOI: 10.1609/aaai.v40i11.37876 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Qixin Sun und 8 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Retrieval Augmented Generation enhances the response accuracy of Large Language Models (LLMs) by integrating retrieval and generation modules with external knowledge, demonstrating particular strength in real-time queries and Visual Question Answering tasks. However, the effectiveness of RAG is frequently hindered by the precision of the retriever: many retrieved samples fed into the generation phase are irrelevant or misleading, posing a critical bottleneck to LLMs’ performance. To address this challenge, we introduce \textbf{VaccineRAG}, a novel Chain-of-Thought-based retrieval-augmented generation dataset. On one hand, VaccineRAG employs a benchmark to evaluate models using data with varying positive/negative sample ratios, systematically exposing inherent weaknesses in current LLMs. On the other hand, it enhances models’ sample-discrimination capabilities by prompting LLMs to generate explicit Chain-of-Thought (CoT) analysis for each sample before producing final answers. Furthermore, to enhance the model’s ability to learn long-sequence complex CoT content, we propose \textbf{Partial-GRPO}. By modeling the outputs of LLMs as multiple components rather than a single whole, our model can make more informed preference selections for complex sequences, thereby enhancing its capacity to learn complex CoT. Comprehensive evaluations and ablation studies on VaccineRAG validate the effectiveness of the proposed scheme.

Zitieren

APA

Sun, Q., Wang, Z., Zhao, H., Li, Y., Song, K., Liu, S., Hu, X., Guo, Q., & Huang, L. (2026). VaccineRAG: Boosting Multimodal Large Language Models’ Immunity to Harmful RAG Samples. Proceedings of the AAAI Conference on Artificial Intelligence, 40(11), 9189-9197. https://doi.org/10.1609/aaai.v40i11.37876

Vancouver

Sun Q, Wang Z, Zhao H, Li Y, Song K, Liu S, et al. VaccineRAG: Boosting Multimodal Large Language Models’ Immunity to Harmful RAG Samples. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(11):9189-9197. doi:10.1609/aaai.v40i11.37876

BibTeX
@article{Sun2026,
  author = {Sun, Qixin and Wang, Ziqin and Zhao, Hengyuan and Li, Yilin and Song, Kaiyou and Liu, Si and Hu, Xiaolin and Guo, Qingpei and Huang, Linjiang},
  title = {VaccineRAG: Boosting Multimodal Large Language Models’ Immunity to Harmful RAG Samples},
  journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
  year = {2026},
  volume = {40},
  number = {11},
  pages = {9189--9197},
  doi = {10.1609/aaai.v40i11.37876}
}
Kostenlos starten →