Prof. Hongmin Cai

South China University of Technology, China

Areas of Professional Interest

Biomedical Image segmentation and posterior analysis, feature selection for bio-data, bioinformatics, big data integration

Speech title：Tensor Spectral Clustering for High-dimension-low-sample-size Data Clustering

Abstract

The need to cluster high-dimensional-low-sample-size (HDLSS) data arise in many scientific fields, like bioinformatics and computer vision. Here, the central problem lies in how to address the noise contamination and concentration effect brought by the high-dimensionality of data. This talk introduces the tensor spectral clustering to address the above issues. First, this approach builds on the tensor similarity framework that characterizes a high-order relation of samples. Specifically, a fourth-order tensor is utilized to depict pair-to-pair similarities, comprehensively characterizes the spatial structure of samples, and thus provides a robust similarity estimation. By incorporating the tensor spectral analysis, we extract the eigenmatrix from the tensor and use it for subsequent clustering. Second, considering the concentration effect caused by the adoption of the Euclidean distance for HDLSS data, we design a discriminating tensor similarity with the anchor-based distance. Through an asymptotic analysis, we prove our discriminating tensor similarity overcomes the concentration effect, leading to an improved clustering performance on HDLSS data. We hope the proposed tensor spectral clustering sheds light on HDLSS data clustering, and provides some insight to machine learning community.

Biography

Dr. Hongmin Cai is a Professor in School of Computer Science and Engineering, South China University of Technology. He is a Senior Member of IEEE, and a Senior Member of China Computer Federation (CCF). He is chair of the conference ICBBB 2021/2022, and serves as editorial members for several journals. He received bachelor and master’s degree from Harbin Institute of Technology in 2001 and 2003, respectively. He got Ph.D from University of Hong Kong in 2007. He was a research fellow at the Center of Bioinformatics at Harvard University, and University of Pennsylvania. In 2012, he was associate Professor of School of Computer Science and Technology, South China University of Technology. In 2014, he was awarded Outstanding Youth Teacher in Guangdong Province. He was promoted to be full professor in 2016. He was a Senior Guest Professor at Kyoto University, Japan. He has principled and co-principled more than 10 fundings from NSF, Guangdong Province and other sources. He has achieved a series of innovative applications in medical imaging and genomics data analysis and made full coverage of famous publications in three fields: artificial intelligence, medical imaging, and bioinformatics. Over 100 research papers have been published with over 50 paper been the first/corresponding author, including IEEE T-PAMI, IEEE T-CYB, IEEE T-IP, IEEE T-MI, Bioinformatics, Briefings in Bioinformatics. His research interests include artificial intelligence, biomedical image processing and bioinformatics