publications

Selected research publications.

CodeBind multimodal alignment overview

CodeBind: Decoupled Representation Learning for Multimodal Alignment with Unified Compositional Codebook

Zeyu Chen, Jie Li, and Kai Han

Findings of the 64th Annual Meeting of the Association for Computational Linguistics (Findings of ACL 2026)

Multimodal representation alignment is essential for large language models and robotics, but existing approaches are limited by cross-modal discrepancies, scarce paired data, and an overemphasis on modality-shared information. CodeBind learns a compositional representation space with shared codebooks for cross-modal semantics and modality-specific details. By incrementally aligning target and bridging modalities, it avoids requiring fully paired data and achieves state-of-the-art classification and retrieval results across nine modalities.

AI in oral health surveillance overview

AI in Oral Health Surveillance: Critical Review

Zeyu Chen, Pei Liu, Kai Han, Peixi Liao, Yanqi Yang, May Chun Mei Wong, Cynthia Kar Yung Yiu, and Edward Chin Man Lo

Journal of Dental Research (JDR)

This critical review examines how artificial intelligence can strengthen oral health surveillance through population-level analysis, intraoral image-based remote screening of oral conditions, and multimodal data integration. It compares the roles of machine learning, computer vision, and emerging multimodal large language models; identifies practical challenges such as imaging quality, domain shift, class imbalance, and cost-effectiveness; and outlines a path toward ethical, scalable, and actionable public-health monitoring systems.