Maijunxian Wang

I work on video reasoning and the evaluation of vision-language models. I build benchmarks and run large-scale experiments to measure how these models reason about the visual world — and where they fail. 我研究视频推理与视觉语言模型的评测,通过构建基准、做大规模实验,衡量这些模型如何理解和推理视觉世界、又在哪里出错。

I'm also interested in Physical AI, world models, and making AI more trustworthy and interpretable. I'm part of VBVR and Grow AI. 我也关注 Physical AI、世界模型,以及让 AI 更可信、更可解释。我是 VBVR 与 Grow AI 的一员。

Selected Publications代表论文

^* equal contribution · ^† equal advising · ^✉ corresponding author · ✓ accepted ^* 共同贡献 · ^† 共同指导 · ^✉ 通讯作者 · ✓ 已录用

A Very Big Video Reasoning Suite (VBVR)

ICML 2026

🤗 Hugging Face #1 Paper of the Month · Feb 2026Hugging Face 当月第一论文 · 2026 年 2 月

Maijunxian Wang^*, Ruisi Wang^*, Juyi Lin^*, Ran Ji^*, Thaddäus Wiedemer, Qingying Gao, Dezhi Luo, Yaoyao Qian, Lianyu Huang, Zelong Hong, Jiahui Ge, Qianli Ma, Hang He, Yifan Zhou, Lingzi Guo, Lantao Mei, Jiachen Li, Hanwen Xing, Tianqi Zhao, Fengyuan Yu, Weihang Xiao, Yizheng Jiao, Jianheng Hou, Danyang Zhang, Pengcheng Xu, Boyang Zhong, Zehong Zhao, Gaoyun Fang, John Kitaoka, Yile Xu, Hua Xu, Kenton Blacutt, Tin Nguyen, Siyuan Song, Haoran Sun, Shaoyue Wen, Linyang He, Runming Wang, Yanzhi Wang, Mengyue Yang, Ziqiao Ma, Raphaël Millière, Freda Shi, Nuno Vasconcelos, Daniel Khashabi, Alan Yuille, Yilun Du, Ziming Liu, Bo Li, Dahua Lin, Ziwei Liu, Vikash Kumar, Yijiang Li, Lei Yang, Zhongang Cai^✉, Hokin Deng^✉

arXiv· Website· EvalKit· Data· Benchmark· Leaderboard· Model

Demystifying Video Reasoning

ECCV 2026

🤗 Hugging Face #2 Paper of the Month · March 2026Hugging Face 当月第二论文 · 2026 年 3 月

Ruisi Wang, Zhongang Cai^✉, Fanyi Pu, Junxiang Xu, Wanqi Yin, Maijunxian Wang, Ran Ji, Chenyang Gu, Bo Li, Ziqi Huang, Hokin Deng, Dahua Lin, Ziwei Liu, Lei Yang

arXiv· Homepage

Vision Language Models Cannot Reason About Physical Transformation (ConservationBench)

ICML 2026

Dezhi Luo^*✉, Yijiang Li^*✉, Maijunxian Wang, Tianwei Zhao, Bingyang Wang, Siheng Wang, Pinyuan Feng, Pooyan Rahmanzadehgervi, Ziqiao Ma, Hokin Deng

arXiv

Egocentric Bias in Vision-Language Models (FlipSet)

CogSci 2026

Maijunxian Wang^*✉, Yijiang Li^*, Bingyang Wang, Tianwei Zhao, Ran Ji, Qingying Gao^†, Emmy Liu^†, Hokin Deng^†, Dezhi Luo^✉

arXiv

Assessing Perceptual Metacognition in Vision-Language Models

CogSci 2026

Ran Ji^*✉, Maijunxian Wang^*, Qingying Gao, Yijiang Li, Hokin Deng, Dezhi Luo^✉

Increasing Computation Resolves Conflicts in Vision Language Models

MMRAgI @ CVPR 2026

Bingyang Wang^*✉, Yijiang Li^*✉, Yitong Qiao, Maijunxian Wang, Tianwei Zhao, Yucheng Sun, Binyue Deng, Hokin Deng, Nuno Vasconcelos, Dezhi Luo^✉

arXiv

Probing Perceptual Constancy in Large Vision-Language Models

ES-Reasoning @ ICLR 2026

Haoran Sun^✉, Bingyang Wang, Suyang Yu, Yijiang Li, Qingying Gao, Haiyun Lyu, Lianyu Huang, Zelong Hong, Jiahui Ge, Qianli Ma, Hang He, Yifan Zhou, Lingzi Guo, Lantao Mei, Maijunxian Wang, Dezhi Luo, Hokin Deng

arXiv

AGI as Second Being: The Structural-Generative Ontology of Intelligence

Preprint预印本

"True intelligence exists only when a system can generate new structures, coordinate them into reasons, and sustain its identity over time."「真正的智能,只存在于一个系统能够生成新结构、将其协调为理由、并在时间中维持其同一性之时。」

Maijunxian Wang, Ran Ji

arXiv

Research Interests研究兴趣

Video reasoning视频推理 Vision-language models视觉语言模型 Benchmarks & evaluation基准与评测 Physical AI World models世界模型 Interpretability可解释性 AI safetyAI 安全