MLLMs Hallucinate when Information Distribution Drifts in Synergy Heads
Meng’en Qin, Junye Chen, Jucheng Liu, Yinchen Liu, Youlu Xing, Song Wang, Ruize Han†
arXiv:2609.09206, 2026
Meng’en Qin, Junye Chen, Jucheng Liu, Yinchen Liu, Youlu Xing, Song Wang, Ruize Han†
arXiv:2609.09206, 2026
Meng’en Qin, Yinchen Liu, Mingxuan Cui, Youlu Xing†
arXiv:2609.19122, 2026
Meng’en Qin*, Yu Song*, Jing Fang, Deyu Meng, Xiaohui Yang†, Lili Yang†
revision resubmitted to TMM, 2026
Meng’en Qin, Yu Song, Quanling Zhao, Xiaodong Yang, Yingtao Che, Xiaohui Yang†
Pattern Recognition, 2026

In this work, we identify three critical issues (information loss, context-agnostic sampling and pattern inconsistency) in existing multi-scale feature frameworks and fusion operations; and propose the theoretically supported A\(^3\)-FPN to optimize multi-scale visual representation learning via the asymptotically disentangled design and content-aware attention modules.
Meng’en Qin, Zhe Li, Hui Huang, Xihong Liu†
arXiv:2604.03337, 2026
Quanling Zhao*, Meng’en Qin*, Yanfeng Sun, Yuan Miao†, Xiaohui Yang†
arXiv:2604.18632, 2025
Lanfeng Wang*, Yanhui Guo*, Zelin Zhang, Meng’en Qin, Zixin Li, Xiaoli Sun, Xiaohui Yang†
MATCH Communications in Mathematical and in Computer Chemistry, 2025
Yanhui Guo*, Lichao Peng*, Zixin Li, Meng’en Qin, Xue Jiao, Yun Chai, and Xiaohui Yang†
MATCH Communications in Mathematical and in Computer Chemistry, 2024