Model merging
M-LossUnderstanding model compatibility and combining trained models with limited unlabeled data.
I am a PhD candidate at Nanyang Technological University (NTU), supervised by Wei Yang Bryan Lim and Chunyan Miao. My research focuses on collaborative AI and model merging.
I am affiliated with the Alibaba–NTU Global e-Sustainability CorpLab (ANGEL). Previously, I was affiliated with the Joint NTU–WeBank Research Centre on Fintech. I received my BSc in Mathematics and Applied Mathematics from The Chinese University of Hong Kong, Shenzhen, in 2023.
Understanding model compatibility and combining trained models with limited unlabeled data.
Studying stability and accelerated optimization in federated learning.
Exploring inference-time pruning and adapter tuning for efficient, secure on-device inference.
Proceedings of the AAAI Conference on Artificial Intelligence, 40(31): 26471–26479, 2026.
@inproceedings{Wang2026MLoss,
title = {{M-Loss: Quantifying Model Merging Compatibility with Limited Unlabeled Data}},
author = {Tiantong Wang and Yiyang Duan and Haoyu Chen and Tiantong Wu and Wei Yang Bryan Lim},
booktitle = {Proceedings of the AAAI Conference on Artificial Intelligence},
year = {2026},
volume = {40},
number = {31},
pages = {26471--26479},
doi = {10.1609/aaai.v40i31.39854},
url = {https://ojs.aaai.org/index.php/AAAI/article/view/39854}
}International Conference on Machine Learning, 2026.
@inproceedings{Gong2026SubspacePath,
title = {{SubspacePath Pruner: Inference-time Pruning via Probe-based Representation--Parameter Coupling}},
author = {Zhiren Gong and Yikun Hou and Fan Wu and Che Wang and Fuyao Zhang and Tiantong Wu and Yurong Hao and Jiaming Zhang and Yiyang Duan and Tiantong Wang and Fei Huang and Chau Yuen and Wei Yang Bryan Lim},
booktitle = {International Conference on Machine Learning},
year = {2026},
url = {https://icml.cc/virtual/2026/poster/62512}
}Advances in Neural Information Processing Systems, 2026.
@inproceedings{Chen2026Stability,
title = {{Stability-Enhanced Federated Learning with Accelerated Gradient}},
author = {Tianxiang Chen and Wenjie Hou and Feng Wang and Tiantong Wang and Zhiming Zheng and Shaoting Tang and Wei Yang Bryan Lim},
booktitle = {Advances in Neural Information Processing Systems},
year = {2026},
url = {https://openreview.net/forum?id=SFp0bLpWTb}
}Advances in Neural Information Processing Systems, 38: 166384–166412, 2025.
@inproceedings{Wang2025AegisGuard,
title = {{AegisGuard: RL-Guided Adapter Tuning for TEE-Based Efficient \& Secure On-Device Inference}},
author = {Che Wang and Ziqi Zhang and Yinggui Wang and Tiantong Wang and Yurong Hao and Jianbo Gao and Tao Wei and Yang Cao and Zhong Chen and Wei Yang Bryan Lim},
booktitle = {Advances in Neural Information Processing Systems},
year = {2025},
volume = {38},
pages = {166384--166412},
url = {https://papers.nips.cc/paper_files/paper/2025/hash/f308b5f207348484552997c536375654-Abstract-Conference.html}
}Neurocomputing, 619: 129053, 2025.
@article{Huang2025AITwins,
title = {{Artificial intelligence without restriction surpassing human intelligence with probability one: Theoretical insight into secrets of the brain with AI twins of the brain}},
author = {Guang-Bin Huang and M. Brandon Westover and Eng-King Tan and Haibo Wang and Dongshun Cui and Wei-Ying Ma and Tiantong Wang and Qi He and Haikun Wei and Ning Wang and Qiyuan Tian and Kwok-Yan Lam and Xin Yao and Tien Yin Wong},
journal = {Neurocomputing},
year = {2025},
volume = {619},
pages = {129053},
doi = {10.1016/j.neucom.2024.129053},
url = {https://doi.org/10.1016/j.neucom.2024.129053}
}Aug 2023 – Present
PhD · College of Computing and Data Science
Supervised by Wei Yang Bryan Lim and Chunyan Miao.
Sep 2019 – Jun 2023
BSc · Mathematics and Applied Mathematics
Applied Mathematics stream · GPA: 3.794 / 4.000.
Academic Performance Scholarship; ranked 1st among 551 students in the School of Science and Engineering in 2019–2020.
Jul 2022
Online Summer Research Program in Physics
Studied particle-spring dynamics with velocity-dependent friction, supervised by Dr. Joao Rodrigues.
Full scholarship co-funded by Cambridge and CUHKSZ.
Sep 2020 – Jun 2023 · CUHKSZ
I delivered weekly tutorials, held office hours, and graded assignments for courses in mathematics, statistics, and physics.
College of Computing and Data Science
Nanyang Technological University
Singapore