Homepage

Hi,my name is Haoyuan Liang (Chinese: 梁浩源).I am a third-year Master’s student at the School of Artificial Intelligence, Sun Yat-sen University. Before that, I obtained my Bachelor’s degree from the School of Mechatronic Engineering and Automation,Shanghai University. I am very fortunate to be advised by Prof. Juepeng Zheng at the Intelligent Geospatial Computing and Sensing (IGCS) Lab.

📖 Research

My research interests are broad, spanning several areas in artificial intelligence and large Language model training. In particular, I focus on the following topics:

⭐ Federated Learning— including domain generalization, and personalized aggregation strategies.
Multimodal Representation Learning — exploring cross-modal alignment and knowledge fusion between vision, language, and audio modalities.
Large Language Model Fine-Tuning — with an emphasis on parameter-efficient adaptation and optimization of foundation models.
Mixture-of-Experts (MoE) Architectures — investigating modular routing, expert specialization, and distributed training efficiency.
Foundation Models for Remote Sensing — building large-scale geospatial pretraining frameworks for earth observation and environmental applications.

❤️ Hobby

Outside of research, I’m also a fitness enthusiast 🏋️‍♂️ and basketball lover 🏀. Feel free to reach out if you’d like to work out together or play a game — I’m always up for some healthy competition and training! 💪
💬 In addition, I serve as an emotional consultant, and I’m always open to sharing insights or offering advice if you ever need to talk about relationships or personal growth. ❤️

📩 If you have an exciting idea or would like to collaborate on my research topics, feel free to reach out via email or WeChat — I’m always open to discussion and new collaborations.
And beyond research, if you’d like to chat about fitness or basketball, or even share and talk about life and relationships, you’re more than welcome to connect with me! 🌟️

🔥 News

2026.01: 🎉🎉 Two papers were accepted by ICLR’25 !
2025.09: 🎉🎉 One paper was accepted by NeurIPS’25 !
2025.08: 🎉🎉 One paper was accepted by EMNLP’25 !
2025.05: 🎉🎉 One paper was accepted by MM’25.
2025.03: 🎉🎉 One paper was accepted by ICME’25 !
2025.03: 🎉🎉 One paper was accepted by CVPR’25.
2024.12: 🎉🎉 One paper was accepted by AAAI’25 !

📝 Publications

: SPFL: Sequential updates with Parallel aggregation for Enhanced Federated Learning under Category and Domain Shifts; Haoyuan Liang, Shilei Cao, Guowen Li, Zhiyu Ye, Haohuan Fu, Juepeng Zheng.; The Thirty-Ninth Annual Conference on Neural Information Processing Systems (NeurIPS'25), 2025, CCF-A

: TTA-FedDG: Leveraging Test-Time Adaptation to Address Federated Domain Generalization; Haoyuan Liang, Xinyu Zhang, Shilei Cao, Guowen Li, Juepeng Zheng.; The Fortieth AAAI Conference on Artificial Intelligence (AAAI'2025), 2025, CCF-A

: Federated Open-Set Domain Generalization with Adaptive Adjustment Boundary and Weights; Haoyuan Liang, Shilei Cao, Yushan Lai, Juepeng Zheng.; IEEE International Conference on Multimedia & Expo 2025 (ICME'25), 2025, CCF-B (Oral)

Other Publication

Yushan Lai, Guowen Li, Haoyuan Liang, Juepeng Zheng, Zhiyu Ye. ADU: Adaptive Detection of Unknown Categories in Black-Box Domain Adaptation. The Thirty-Sixth IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR’25), 2025 (CCF-A)
Zhiyu Ye, Guowen Li, Haoyuan Liang, Zixi Wang, Shilei Cao, Yushan Lai, Juepeng Zheng. Quantifying Samples with Invariance for Source-Free Class Incremental Domain Adaptation. the 33rd ACM International Conference on Multimedia (MM’25) (CCF-A)
Jielong Tang, Yang Yang, Jianxing Yu, Zhen-Xing Wang, Haoyuan Liang, Liang Yao, Jian Yin. UnCo: Uncertainty-Driven Collaborative Framework of Large and Small Models for Grounded Multimodal NER. The 2025 Conference on Empirical Methods in Natural Language Processing (EMNLP’25) (CCF-B)

🙂 Preprint

Haoyuan Liang, Zhiyu Ye, Jielong Tang, Yang Yang, Shilei Cao, Fei Hu, Guowen Li, Zhiwei Zhang, Haohuan Fu, Juepeng Zheng. UniFLoW: Universal Multi-Modal Federated LoRA Fine-Tuning Framework with Analytical Aggregation. Submitted to The Fourteenth International Conference on Learning Representations (ICML’2026) (CCF-A)
Juepeng Zheng, Haoyuan Liang, Shilei Cao, Qingmei Li, Yushan Lai, Guowen Li, Haohuan Fu. Federated Open-Set Domain Generalization with Adaptive Adjustment Boundary and Weights. Submitted to The IEEE Transactions on Neural Networks and Learning Systems (TNNLS) (CCF-B)

🏆 Honor & Award

🎓 Graduate Period (Sun Yat-sen University)

2025.09 National Scholarship (国家奖学金), Sun Yat-sen University
2025.09 First-Class Academic Scholarship (学业一等奖学金), Sun Yat-sen University
2025.08 Huawei Innovation Cup – National Final, Research Excellence Award (鲲鹏创新应用大赛全国总决赛科研赛道银奖)
2025.06 First Prize, 1st National Land Surface Resource Management GIS Application Competition (全国陆表资源管理地理信息技术应用竞赛一等奖)

🎓 Undergraduate Period (Shanghai University)

2023.06 Outstanding Graduate (优秀毕业生), Shanghai University
2022.12 National Scholarship (国家奖学金), Shanghai University
2022.08 MCM/ICM – M Award (美国大学生数学建模竞赛 M 奖)
2021.12 National Second Prize, China Undergraduate Mathematical Contest in Modeling (全国大学生数学建模竞赛国家级二等奖)
2021.11 Second Prize, National Advanced Manufacturing Product Innovation Competition (全国大学生先进制造技术与产品信息建模创新大赛国家二等奖)
🏀 Runner-up, Three-Point Shooting Contest (三分球大赛亚军), School of Mechatronic Engineering and Automation, Shanghai University

Haoyuan Liang(梁浩源)