Welcome to KAWAII AI lab @ HKU!

Hi! KAWAII AI lab (Knowledge Acquisition With Artificial Intelligence Innovation laboratory) at Department of Mathematics, The University of Hong Kong (HKU), led by Professor SUZUKI, Atsushi, aims to enhance human knowledge through artificial intelligence (AI) research.

Our main research interests are

• Theoretical (e.g., statistical) behavior of AI algorithms,
• Knowledge discovery through the lens of AI, and
• Discovering novel usages and applications of AI,

with the helps of math.

Our research interests are NOT bound to any particular data domain BUT rather covering various data domains.

As research tools, we mainly use mathematics, such as statistics, information theory, learning theory, Riemannian geometry, etc.

News

• Our paper, "Fundamental Limitation in Explaining AI," has been uploaded to arXiv (2026-05-23).
• Our paper, "Normalized Maximum Likelihood Code-Length on Riemannian Manifold Data Spaces," has been published in IEEE Transactions on Information Theory (2026-05-08). (arXiv)
• Our student, XIAO, Suyang, has been selected as one of the best poster presenters in Research Colloquium for Science Undergraduate Students 2025-26, The University of Hong Kong. Congratulations, Suyang!
• Our paper, "Overfitting has a limitation: a model-independent generalization gap bound based on Rényi entropy," has been uploaded to arXiv (2025-05-30).
• Official X account has launched (2025-02-27).
• Our paper, "Hallucinations are inevitable but can be statistically negligible," has been uploaded to arXiv (2025-02-15). Brief introduction on X
• The group has moved to Department of Mathematics, The University of Hong Kong (HKU) (2025-01-01).
• Our paper, "Foundation of Calculating Normalized Maximum Likelihood for Continuous Probability Models," has been uploaded to arXiv (2024-09-12). Brief introduction on X
• The new website launched (2024-09-03).