.png?sfvrsn=d3a97fa3_12)
News
Features
3 Papers Accepted at EMNLP 2025
10 Oct 2025
As one of the premier conferences in natural language processing, the 2025 Conference on Empirical Methods in Natural Language Processing (EMNLP 2025) will take place in Suzhou, China, from 4 to 9 November. We are proud to share that three papers from A*STAR Centre for Frontier AI Research (A*STAR CFAR) have been accepted at this prestigious conference.
Congratulations to the following scientists:
- Dr Joey Zhou, Deputy Director, Principal Scientist
- Dr Basura Fernando, Principal Scientist
- Dr Du Jiawei, Senior Scientist
List of accepted papers:
| 1. | Diagram-Driven Course Questions Generation Xinyu Zhang*, Lingling Zhang, Yanrui Wu, Muye Huang, Wenjun Wu, Bo Li, Shaowei Wang, Basura Fernando, Jun Liu Visual Question Generation (VQG) research has largely focused on natural images while overlooking diagrams, which are critical in educational materials. To address this gap, the authors propose the Diagram-Driven Course Questions Generation (DDCQG) task and introduce DiagramQG, a comprehensive dataset containing 15,720 diagrams and 25,798 questions across 37 subjects and 371 courses, supporting more pedagogically relevant question generation. |
| 2. | RRInf: Efficient Influence Function Estimation via Ridge Regression for Large Language Models and Text-to-Image Diffusion Models Zhuozhuo Tu, Cheng Chen*, Yuxuan Du RRInf introduces a principled method for influence function estimation in large-scale generative AI models by reformulating the problem as ridge regression. This insight enables the development of an algorithm that is both efficient and scalable to modern LLMs and diffusion models. |
| 3. | Agent Trading Arena: A Study on Numerical Understanding in LLM-Based Agents Tianmi Ma, Jiawei Du, Wenxin Huang, Wenjie Wang, Liang Xie, Xian Zhong, Joey Tianyi Zhou The Agent Trading Arena is a competitive multi-agent stock market simulation where LLM-based agents directly affect price dynamics through bid–ask interactions. Experiments show that incorporating chart-based visualisations and a reflection module significantly improves agents’ numerical reasoning and trading performance, particularly under high volatility. |
*denotes current student at A*STAR CFAR
(accurate at the time of posting)
Learn more about EMNLP 2025.
A*STAR celebrates International Women's Day
From groundbreaking discoveries to cutting-edge research, our researchers are empowering the next generation of female science, technology, engineering and mathematics (STEM) leaders.
Get inspired by our #WomeninSTEM