Hi๐! Iโm Xintong Li (ๆๆฌฃๆก). Iโm an undergraduate student in the Honors Talent Program in Statistics and Data Science at Renmin University of China. Iโm very fortunate to be advised by Prof. Xiao Zhou at the LUMOS AI Lab. I also have the incredible opportunity to work with fantastic mentors at Rutgers University. At the ARSENAL Lab, Iโm co-advised by Prof. Desheng Zhang, Prof. Guang Wang, Prof. Zhiqing Hong, and work closely with Zejun Xie. Additionally, Iโm grateful to be advised by Prof. Yifan Hu.
My research interests mainly lie in the intersection of HumanโAI Collaboration, Causal Inference and Discovery, Spatio-temporal Modeling, and AI for Social Good. I am excited about building theoretically grounded frameworks that work robustly in complex, real-world environments.
I am actively seeking Ph.D. or Master's opportunities for Fall 2026! ๐ and am eager to connect with fellow researchers and potential advisors!
๐๏ธ News
2026.01 โ Our paper AtC was accepted to ICLR-26 ๐
๐ See you in Brazil ๐ง๐ท!2025.11 โ Our paper ST-HCM was accepted to AAAI-26!
๐ See you in Singapore ๐ธ๐ฌ!2024.12 โ I received the AAAI-25 Student Scholarship ($2000).
Feeling grateful and excited to attend AAAI again ๐ฅฐ!2024.12 โ Our paper FAP-CD was accepted to AAAI-25.
Looking forward to sharing our work with the community ๐บ๐ธ!2023.09 โ I joined LUMOS AI Lab at the GSAI, advised by Prof. Xiao Zhou.
This marked the beginning of my research journey ๐.
๐ Publications
* denotes co-first authorship.
AtC: Aggregate-then-Calibrate for Human-centered Assessment [PDF] [Code]
Zejun Xie*, Xintong Li*, Guang Wang, and Desheng Zhang.
The Fourteenth International Conference on Learning Representations (ICLR 2026).TL;DR: AtC aggregates human comparisons into a consensus and isotonic-calibrates any modelโs scores via the consensus, delivering decision-ready assessments with guarantees on efficiency, robustness, and optimality.
Spatio-Temporal Hierarchical Causal Models [PDF] [Code] [Poster] [Slides] [Video] Xintong Li, Haoran Zhang, and Xiao Zhou.
In Proceedings of the AAAI Conference on Artificial Intelligence (AAAI 2026).TL;DR: We introduce Spatio-Temporal Hierarchical Causal Models (ST-HCMs) to perform causal inference from spatio-temporal data with unobserved, time-invariant confounders, a setting where traditional methods fail.
FAP-CD: Fairness-Driven Age-Friendly Community Planning via Conditional Diffusion Generation [PDF] [Code] [Poster]
Jinlin Li, Xintong Li, and Xiao Zhou.
In Proceedings of the AAAI Conference on Artificial Intelligence (AAAI 2025).TL;DR: FAP-CD utilizes a conditioned graph denoising diffusion probabilistic model to learn the conditional joint probability distribution of aging facilities and their spatial relationships, and generates optimized spatial distributions of facilities.
๐ Education
Renmin University of China, Beijing, China (Sep. 2022 โ Jul. 2026)
- Bachelor of Science in Statistics
- Honors Talent Program in Statistics and Data Science
โจ Misc.
As a member of the Youth Volunteer Association, I organized the โReach Outโ project, a platform for international students to learn Chinese and explore local culture. ๐ [Intro] [Summary] [Recruitment]
Iโm currently volunteer to help stray cats in my community. ๐๐ฅฐ (Hereโs one of my buddies!)