{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,2]],"date-time":"2026-06-02T09:59:28Z","timestamp":1780394368751,"version":"3.54.1"},"reference-count":0,"publisher":"Association for the Advancement of Artificial Intelligence (AAAI)","issue":"34","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["AAAI"],"abstract":"<jats:p>Multimodal Large Language Models (MLLMs) have played an increasingly important role in multimodal intelligence. However, the existing fine-tuning methods often ignore cross-modal heterogeneity, limiting their full potential. In this work, we propose a novel fine-tuning strategy by injecting beneficial random noise, which outperforms previous methods and even surpasses full fine-tuning, with minimal additional parameters. The proposed Multimodal Noise Generator (MuNG) enables efficient modality fine-tuning by injecting customized noise into the frozen MLLMs. Specifically, we reformulate the reasoning process of MLLMs from a variational inference perspective, upon which we design a multimodal noise generator that dynamically analyzes cross-modal relationships in image-text pairs to generate task adaptive beneficial noise. Injecting this type of noise into the MLLMs effectively suppresses irrelevant semantic components, leading to significantly improved cross-modal representation alignment and enhanced performance on downstream tasks. Experiments on two mainstream MLLMs, QwenVL and LLaVA, demonstrate that our method surpasses full parameter fine-tuning and other existing fine-tuning approaches, while requiring adjustments to only about 1~2% additional parameters.<\/jats:p>","DOI":"10.1609\/aaai.v40i34.40153","type":"journal-article","created":{"date-parts":[[2026,3,18]],"date-time":"2026-03-18T02:29:04Z","timestamp":1773800944000},"page":"29150-29158","source":"Crossref","is-referenced-by-count":2,"title":["Explore How to Inject Beneficial Noise in MLLMs"],"prefix":"10.1609","volume":"40","author":[{"given":"Ruishu","family":"Zhu","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Sida","family":"Huang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ziheng","family":"Jiao","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Hongyuan","family":"Zhang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"9382","published-online":{"date-parts":[[2026,3,14]]},"container-title":["Proceedings of the AAAI Conference on Artificial Intelligence"],"original-title":[],"link":[{"URL":"https:\/\/ojs.aaai.org\/index.php\/AAAI\/article\/download\/40153\/44114","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/ojs.aaai.org\/index.php\/AAAI\/article\/download\/40153\/44114","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,18]],"date-time":"2026-03-18T02:29:04Z","timestamp":1773800944000},"score":1,"resource":{"primary":{"URL":"https:\/\/ojs.aaai.org\/index.php\/AAAI\/article\/view\/40153"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,3,14]]},"references-count":0,"journal-issue":{"issue":"34","published-online":{"date-parts":[[2026,3,17]]}},"URL":"https:\/\/doi.org\/10.1609\/aaai.v40i34.40153","relation":{},"ISSN":["2374-3468","2159-5399"],"issn-type":[{"value":"2374-3468","type":"electronic"},{"value":"2159-5399","type":"print"}],"subject":[],"published":{"date-parts":[[2026,3,14]]}}}