My research interests are in computer vision and machine learning, with a focus on developing ideas that are both academically grounded and impactful in practice. I lead and mentor a research team at Z.UM Lab, Universiti Malaya working across computer vision and machine learning, with research published at leading international venues including CVPR, NeurIPS, ICLR, ACL, and EMNLP. I also had the privilege of serving as the founding Chair of the IEEE Computational Intelligence Society Malaysia Chapter in 2015.
I currently serve as an Associate Editor of Pattern Recognition and IEEE Transactions on Circuits and Systems for Video Technology, and have co-organized numerous conferences, workshops, tutorials, and challenges in computer vision and machine learning. Over the years, my work has been recognized through several honors, including the Malaysia Scopus Research Excellence Award (Research Innovations) in 2024, Top Research Scientists Malaysia (TRSM) in 2022, membership in the Young Scientists Network–Academy of Sciences Malaysia (YSN-ASM) in 2015, and the Hitachi Research Fellowship in 2013. I am also a Senior Member of IEEE and a Professional Engineer (BEM).
Between 2020 and 2022, I was seconded to the Ministry of Science, Technology and Innovation (MOSTI) as Undersecretary for the Division of Data Strategic and Foresight, where I contributed to national efforts in data strategy and future-oriented planning. I remain committed to nurturing the next generation of researchers, supporting Malaysian talent to excel globally, and advancing a responsible and inclusive AI ecosystem that reflects our nation’s values and aspirations.
Highlights:
10/2026: Two(2) papers to appear in NeurIPS-2026
10/2026: One(1) paper to appear in ACCV-2026
10/2026: One(1) paper to appear in AACL-IJCNLP-2026
08/2026: One(1) paper to appear in EMNLP-2026 (main)
04/2026: One(1) paper to appear in ACL-2026 (main, long paper)
02/2026: One(1) paper to appear in CVPR-2026 (main)
01/2026: One(1) paper to appear in ICLR-2026.
professor at Universiti Malaya
Latest Works
Between Multiple Choice and Open Response: Evaluating LLMs with Option-Free Deterministic Scoring 
Z.K. Hew, S.J. Yang and C.S. Chan
EMNLP 2026 (main, acceptance rate: 2719/17,669 ~ 15.4%)
This paper shows that LLM benchmark scores depend on the evaluation instrument and introduces CSRQ as a deterministic middle ground between MCQs and judge-scored open responses, validated on converted benchmarks. Also we built MyCulture, the first Malaysia-centred cultural benchmark in Bahasa Malaysia, with 2,652 questions across six cultural domains.
pdf poster slide 11mins video Project Page MyCulture datasetSingle-Pass, Depth-Selective Reading for Multi-Aspect Sentiment Analysis 
Y. Xia, Z. Pan, A. Kamsin and C.S. Chan
ACL 2026 (main, long paper, acceptance rate: 2308/12,148 ~ 19%)
This paper proposes a single-pass, depth-selective reading framework for multi-aspect sentiment analysis, designed to reduce computational redundancy by separating shared sentence encoding from aspect-specific querying. Our work managed to reduce end-to-end computation by up to 60% in multi-aspect settings (M > 2).
pdf codeOneHOI: Unifying Human-Object Interaction Generation and Editing 
J.T. Hoe, W. Hu, X. Jiang, Y-P. Tan and C.S. Chan
CVPR 2026 (main, acceptance rate: 4090/16,092 ~ 25.42%)
Rather than treating generation and editing as separate problems, this paper - OneHOI models both as a unified conditional denoising process grounded in structured ⟨person, action, object⟩ relations, enabling (i) Single-interaction generation; (ii) Identity-preserving editing, and (iii) Multi-HOI editing. All within one coherent framework.
pdf poster Project Page code HOI-Edit-44K datasetTowards Privacy-Guaranteed Label Unlearning in Vertical Federated Learning: Few-Shot Forgetting without Disclosure 
H. Gu, H.X. Tae, L. Fan and C.S. Chan
ICLR 2026 (acceptance rate: 5339/18,949 ~ 28.18%)
We address label unlearning in Vertical Federated Learning (VFL), where labels are essential for training yet also privacy-sensitive. Our approach uses representation-level manifold mixup to generate synthetic embeddings for both forgotten and retained samples, enabling more effective and efficient gradient-based forgetting while preserving model performance
pdf poster code