Xuan Kan

prof_pic.jpg

I’m Xuan Kan (阚璇), a Senior Research Scientist at Meta Monetization GenAI. I develop creative agents and multimodal evaluation systems for AI-generated advertising, with a focus on reinforcement learning, reward modeling, post-training, and multimodal foundation models.

I earned my Ph.D. in Computer Science from Emory University, advised by Prof. Carl Yang and Prof. Ying Guo. My doctoral research focused on machine learning for brain imaging and neuroinformatics. Previously, I studied Software Engineering at Tongji University and worked on efficient neural networks, pervasive sensing, and trustworthy machine learning at SenseTime and with Prof. Xiaoxuan Lu at the University of Oxford.

news

May 20, 2026 Our work entitled Multi-Task Reinforcement Learning for Enhanced Multimodal LLM-as-a-Judge has been accepted to ACL 2026 as an oral presentation. It was a privilege to mentor my intern Junjie Wu on this project, and I look forward to presenting it in San Diego!
May 18, 2024 I’m excited to announce that I’ve graduated from Emory University with my PhD and have joined Meta Monetization GenAI as a Research Scientist! I want to express my heartfelt gratitude to my advisors, Prof. Carl Yang and Prof. Ying Guo, for their exceptional guidance and support throughout my doctoral journey.
Sep 21, 2023 Our work entitled Open Visual Knowledge Extraction via Relation-Oriented Multimodality Model Prompting accepted to NeurIPS 2023. Congrats to Hejie!
Aug 25, 2023 One paper entitled Dynamic Brain Transformer with Multi-level Attention for Functional Brain Network Analysis has been accepted for IEEE BHI 2023. Very appreciate the travel awrad from the conference. See you in Pittsburgh!
Jun 10, 2023 Ending my Google Intership and will start my next journey at Meta in Seattle, WA. See you soon, Seattle!

selected publications

  1. ACL
    Multi-Task Reinforcement Learning for Enhanced Multimodal LLM-as-a-Judge
    Junjie Wu, Xuan Kan, Zihao He, and 3 more authors
    In The 64th Annual Meeting of the Association for Computational Linguistics, 2026
    Oral arXiv
  2. IEEE BHI
    Multi-task Learning for Brain Network Analysis in the ABCD study
    Xuan Kan, Hejie Cui, Keqi Han, and 2 more authors
    In The IEEE-EMBS International Conference on Biomedical and Health Informatics, 2024
  3. KDD
    R-Mixup: Riemannian Mixup for Biological Networks
    Xuan Kan, Zimu Li, Hejie Cui, and 6 more authors
    Proceedings of the ACM International Conference on Knowledge Discovery and Data Mining, 2023
    arXiv
  4. NeurIPS
    Brain Network Transformer
    Xuan Kan, Wei Dai, Hejie Cui, and 3 more authors
    Proceedings of the Conference on Neural Information Processing Systems, 2022
    arXiv Code Poster
  5. MIDL
    FBNetGen: Task-aware GNN-based fMRI Analysis via Functional Brain Network Generation
    Xuan Kan, Hejie Cui, Joshua Lukemire, and 2 more authors
    Medical Imaging with Deep Learning, 2022
    arXiv Code