Xilong Cheng (Hsi-lung Cheng)

xilong-cheng.png

Xilong Cheng received his Master’s degree from the School of Information and Communication Engineering, Communication University of China, in 2026, under the supervision of Yunxiao Qin. He currently works on multimodal large language models at Mininglamp Technology. He is also a remote visiting scholar with the X-NLP Lab at INSAIT, collaborating closely with Yuxia Wang. He is dedicated to making more intelligent, safer personalized AI accessible to more people.

His research interests include AI personalization, long-term memory, and learning from the real world through egocentric vision.

Publications

Personalization & Memory

  1. TACL
    Illustration for PsyMem: Fine-grained Psychological Alignment and Explicit Memory Control for Advanced Role-Playing LLMs
    PsyMem: Fine-grained Psychological Alignment and Explicit Memory Control for Advanced Role-Playing LLMs
    Xilong Cheng*, Yunxiao Qin*†, Yuting Tan*, Zhengnan Li, Ye Wang, Hongjiang Xiao†, and Yuan Zhang
    Transactions of the Association for Computational Linguistics, 2026
    Paper Code (not yet released) Weights (not yet released)
  2. arXiv
    Illustration for PsyMemBehav: Psychology-Informed and Memory-Grounded Two-Level Identity-to-Action Reasoning for Social-Media Behavior Simulation
    PsyMemBehav: Psychology-Informed and Memory-Grounded Two-Level Identity-to-Action Reasoning for Social-Media Behavior Simulation
    Wenji Zhang*, Xilong Cheng*§, Yunxiao Qin*†, Mai Tu, Tongye Zhu, Yuhao Zhou, JingBo Ji, Yuting Tan, Duoyin Li, Hongjiang Xiao†, and Yuan Zhang
    2026
    arXiv preprint
    Paper Code (not yet released) Weights (not yet released)
  3. arXiv
    Illustration for MemRole: From Memories to Decisions for Long-Horizon Reasoning in Role-Playing Agents
    MemRole: From Memories to Decisions for Long-Horizon Reasoning in Role-Playing Agents
    Mengting Li*, Xilong Cheng*§, Zhihao Xiao, Zhenhua Xu, Jingyuan Zhou, Linfeng Li, Long Zhang, Borui Cai†, and Yunxiao Qin†
    2026
    arXiv preprint
    Paper Code (not yet released) Weights (not yet released)
  4. Illustration for ArcRole: Scene-Grounded Character Growth for Long-Horizon Role-Playing Agents
    ArcRole: Scene-Grounded Character Growth for Long-Horizon Role-Playing Agents
    Xilong Cheng, Zhenhua Xu, Mengting Li, Yuting Tan, Jingyuan Zhou, Yuhao Zhou, JingBo Ji, Wenji Zhang, Duoyin Li, Yunxiao Qin†, Hongjiang Xiao†, and Yuan Zhang
    Manuscript
    Paper (not yet released) Code (not yet released) Weights (not yet released)
  5. Illustration for AURA: Scope-Grounded Evidence Activation for Long-Term Conversational Memory
    AURA: Scope-Grounded Evidence Activation for Long-Term Conversational Memory
    Duoyin Li*, Xilong Cheng*, Ruoyu Chen, Wenji Zhang, Yuhao Zhou, Yuting Tan, and Yunxiao Qin
    Manuscript
    Paper (not yet released) Code (not yet released)
  6. Illustration for HiTaC-Mem: Enhancing Temporal and Causal Reasoning in Grounded LLM Systems with Hierarchical Graph Memory
    HiTaC-Mem: Enhancing Temporal and Causal Reasoning in Grounded LLM Systems with Hierarchical Graph Memory
    Yuhao Zhou, Xilong Cheng, Wenji Zhang, Duoyin Li, JingBo Ji, Yuting Tan, Yunxiao Qin, and Yuan Zhang
    Manuscript
    Paper (not yet released) Code (not yet released)
  7. arXiv
    Illustration for RippleMem: From Isolated Retrieval to Associative Recollection for Long-Term Agent Memory
    RippleMem: From Isolated Retrieval to Associative Recollection for Long-Term Agent Memory
    Jingbo Ji, Lingyi Li, Xilong Cheng, Yuhao Zhou, Wenji Zhang, Yuting Tan, and Yunxiao Qin†
    2026
    arXiv preprint
    Paper Code (not yet released)
  8. Survey
    Illustration for LLM-based Personal Digital Twins: A Survey
    LLM-based Personal Digital Twins: A Survey
    Yuyang Dai*, Yan Lin*, Asen Dotsinski*, Xilong Cheng, Jiayi Fu, Kaiyang Wan, and Yuxia Wang
    Manuscript
    Paper

Egocentric Learning

  1. CVPR 2026
    Illustration for Towards Stable Self-Supervised Object Representations in Unconstrained Egocentric Video
    Towards Stable Self-Supervised Object Representations in Unconstrained Egocentric Video
    Yuting Tan*, Xilong Cheng*, Yunxiao Qin*†, Zhengnan Li, and Jingjing Zhang†
    In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2026
    Paper Code (not yet released) Weights (not yet released)

Other Research

  1. EMNLP Main 2026
    Illustration for SRAF: Stealthy and Robust Adversarial Fingerprint for Copyright Verification of Large Language Models
    SRAF: Stealthy and Robust Adversarial Fingerprint for Copyright Verification of Large Language Models
    Zhebo Wang*, Zhenhua Xu*, Maike Li, Xilong Cheng, Wenpeng Xing, Chunqiang Hu, Chen Zhi, and Meng Han†
    In Proceedings of the Conference on Empirical Methods in Natural Language Processing (Main Conference), 2026
    Paper Code (not yet released)
  2. arXiv
    Illustration for FSMLP: Modelling Channel Dependencies With Simplex Theory Based Multi-Layer Perceptions In Frequency Domain
    FSMLP: Modelling Channel Dependencies With Simplex Theory Based Multi-Layer Perceptions In Frequency Domain
    Zhengnan Li, Haoxuan Li, Hao Wang, Jun Fang, Zhichao Chen, Yuting Tan, Xilong Cheng, and Yunxiao Qin†
    2024
    arXiv preprint
    Paper
  3. ESWA
    Illustration for Meta-Tuner: Meta-Trained Node-Specific Transformations for Graph Few-Shot Class-Incremental Learning
    Meta-Tuner: Meta-Trained Node-Specific Transformations for Graph Few-Shot Class-Incremental Learning
    Zhengnan Li, Jun Fang, Junbo Wang, Xilong Cheng, Yuting Tan, and Yunxiao Qin†
    Expert Systems with Applications, 2025
    Paper
  4. IEEE SPL
    Illustration for FTMixer: Frequency and Time Domain Representations Fusion for Time Series Forecasting
    FTMixer: Frequency and Time Domain Representations Fusion for Time Series Forecasting
    Zhengnan Li, Yuting Tan, Xilong Cheng, and Yunxiao Qin†
    IEEE Signal Processing Letters
    Paper

* Equal contribution · † Corresponding author · § Project leader. Gray buttons indicate resources that have not yet been released. I plan to release resources for all accepted papers together as soon as possible. My recent heavy workload has delayed these releases.

Experience

Large Language Model Algorithm Engineer

Mininglamp Technology · Full-time

Reporting directly to @Chenxu Zhao (VP).

– Present

Post-training and capability optimization for Computer Use Agents (CUA).

Large Language Model Algorithm Engineer

Alibaba Group · Internship

–

Continual pre-training (CPT) of large language models for applications.

Research Assistant

Chinese Academy of Sciences · Internship

–

Algorithm optimization.

Education

  • 🎓 M.S. in Information & Communication Engineering, 2023–2026
    Communication University of China · Advisor: Yunxiao Qin
  • 🎓 B.S. in Information & Communication Engineering, 2018–2022
    Communication University of China

Interests

Beyond my academic focus on Large Language Models and Role-Playing Language Agents, I also have a strong passion for sports, particularly football and, more recently, tennis. While I see sports as an essential form of relaxation, I equally enjoy streaming media and reading literature on modern psychology, which help me unwind and broaden my perspective.