| 标题 |
 Dynamic hierarchical memory improved mixture-of-experts for multimodal fake news detection 用于多模态假新闻检测的动态分层记忆改进专家混合
相关领域
计算机科学
假新闻
人工智能
特征（语言学）
钥匙（锁）
动态随机存取存储器
计算机视觉
噪音（视频）
模式识别（心理学）
遮罩（插图）
|
| 网址 | |
| DOI | |
| 其它 |
Abstract The explosive growth of social media information has exacerbated the risk of fake news dissemination. This paper proposes a Dynamic Hierarchical Memory Improved Mixture-of-Experts for Multimodal Fake News Detection, named MemiMoE-FND. Specifically, MemiMoE-FND first utilizes advanced feature encoders to extract multi-scale feature representations of modalities. Then, the Memory Improved Mixture-of-Experts is introduced to enhance the feature refinement ability. Next, the Dynamic Hierarchical Memory Improved Mixture-of-Experts is designed, using a two-level expert network to progressively refine and fuse features, achieving collaborative optimization of multi-scale features. Finally, a lightweight Fusion_Gate is designed to dynamically adjust feature contributions while ensuring transparency in the decision-making process. Experimental results on three benchmark datasets |
| 求助人 | |
| 下载 | |
|
温馨提示：该文献已被科研通 学术中心 收录，前往查看
科研通『学术中心』是文献索引库，收集文献的基本信息（如标题、摘要、期刊、作者、被引量等），不提供下载功能。如需下载文献全文，请通过文献求助获取。
|
PDF的下载单位、IP信息已删除
Criminology34
ASD
David
GingerF
你嵙这个期刊没买
竹筏过海
寻道图强
tuanheqi
优美的小笨蛋
嘻嘻哈哈
elliotzzz
从容芮
ccm
erg
Mollyxueyue
刻苦的念柏
avalanche
mmyhn
无极微光
9527
Progie
xxfsx
franklin_fsz
皖公网安备34019202002308
