XMUM Undergraduate Research Accepted to Top International AI Conference
2025-10-02Xiamen University Malaysia (XMUM) congratulates Fang Zitao, Yu Shuyang, Guo Yifei, Zhang Yiwei, Cao Yiyao, and Dr. Sim Kuan Goh from the School of Artificial Intelligence and Robotics (SAIR) on their paper acceptance at the 2025 Conference on Empirical Methods in Natural Language Processing (EMNLP 2025).
Recognized as a leading international conference in Natural Language Processing (NLP), EMNLP is ranked A* by the Computing Research and Education (CORE) Conference Ranking.
Their article, entitled "To See a World in a Spark of Neuron: Disentangling Multi-task Interference for Training-free Model Merging", revisits the role of neurons in fine-tuning across vision, language, and large language models (LLMs), and introduces NeuroMerging, a novel approach that provides a fresh neuronal perspective on unifying multiple AI models with complementary strengths into a single multi-task model. This is particularly significant with the rapid growth of publicly shared AI models (e.g., on Hugging Face), as NeuroMerging can fully leverage and integrate their complementary strengths without incurring additional training or inference costs.
This article is first-authored by Fang Zitao, a final-year undergraduate student from Artificial Intelligence Programme, with Dr. Sim Kuan Goh as the supervisor.
Yu Shuyang and Guo Yifei are both Class of 2025 graduate, who are currently pursuing Master's degrees at Columbia University and Duke University, respectively. The research also benefited from the collaborative contributions of Guodong Du, Jing Li, and Ho-Kin Tang from the Hong Kong Polytechnic University and the Harbin Institute of Technology, Shenzhen.
This work was supported in part by the Ministry of Higher Education Malaysia through the Fundamental Research Grant Scheme, the National Science Foundation of China, the Shenzhen Science and Technology Program, the Department of Science and Technology of Guangdong, Shenzhen Start-Up Research Funds, and the Xiamen University Malaysia Research Fund.
Read the preprint on ArXiv: https://arxiv.org/abs/2503.05320
Explore more on the project page: https://zzzitaofang.github.io/projects/NeuroMerging/
| Fang Zitao is currently a final-year undergraduate student at Xiamen University Malaysia, under the supervision of Dr. Goh Sim Kuan. His research topic of interest focuses on knowledge fusion, alignment, and compression of generative models, particularly large language models, as well as representation learning and self-supervised learning for multivariate time-series data. |
| Sim Kuan Goh is currently an Assistant Professor at the School of Artificial Intelligence and Robotics, Xiamen University Malaysia. He holds a BEng and PhD from the National University of Singapore. His research interests are in Multimodal AI and Brain-Computer Interface, and their applications. He is an IEEE Senior Member and has shared his work in journals and conferences (e.g., NeurIPS, ACL, EMNLP, and others). |
