Tutorial Speakers
Likeng Liang
South China Normal University
Dr. Likeng Liang is a Postdoctor at the School of Computer Science, South China Normal University, Guangzhou, China. His research fields include medical informatics, natural language processing, clinical research. He has published more than 10 papers on journals or conferences such as Drug Safety, Neural Computing and Applications, BMC Medical Informatics & Decision Making. He is also a PC member of many medical informatics conferences such as NCAA, CHIP, ICIC, HealthBDAI.
Tutorial Title
Medical Concept Normalization Based on Neural Network
Tutorial Abstract
With the rapid growth of biomedical literature, concepts generally have various mentions owing to user preference, culture, or other factors. Such problem urgently needs to be solved for information retrieval, information extraction, health statistics, etc. Medical concept normalization (MCN) aims to solved this problem by constructing semantic mapping between concepts and mentions. In recent years, neural network--based methods have shown enormous potential and achieved exciting progress in MCN. These methods can learn semantic mapping from the real data and is robust to complicated input. This tutorial aims to provide the audience with a general picture of neural network--based MCN. Specifically, we will introduce the algorithm framework of MCN, review the most important works and discuss MCN in large-scale databases. Future research directions and challenges are also pointed out.
Jicong Fan
The Chinese University of Hong Kong, Shenzhen
Dr. Jicong Fan is a Research Assistant Professor at the School of Data Science, The Chinese University of Hong Kong, Shenzhen, China. Prior to joining the institute, he was a Postdoctoral Associate at the School of Operations Research and Information Engineering, Cornell University, Ithaca, USA. He held research positions at The University of Wisconsin-Madison and The University of Hong Kong in 2018 and 2015 respectively. He obtained his Ph.D. from the Department of Electronic Engineering, City University of Hong Kong in 2018. Dr. Fan's research fields include machine learning, computer vision, optimization, neuroscience signal processing, and statistical process control. He has, as the first author or corresponding author, published more than 20 papers on prestigious journals and conferences such as IEEE TSP, KDD, NeurIPS, ICLR, CVPR, and AAAI. He is serving as a reviewer of many journals such as IEEE TNNLS/TII/TIE/TIP/TSP/TKDE, Pattern Recognition, Signal Processing, SIAM Journal on Mathematics of Data Science, and Mathematical Programming. He is also a PC member of many top AI conferences such as COLT, AISTATS, NeurIPS, CVPR, AAAI, KDD, and ICML.
Tutorial Title
Recent Advances of Missing Data Imputation
Tutorial Abstract
Missing data are prevalent in science and engineering. For example, in bioinformatics, single-cell gene data are often incomplete. In recommendation system, the rating matrices are always highly incomplete. In computer vision, images and videos may have missing pixels or regions due to noise or occlusion. In electronic or chemical engineering, the data may have missing values because of the failure of sensors. The problem of missing data imputation has been studied for more than fifty years and many missing data imputation methods have been proposed. Perhaps low-rank matrix factorization and deep neural networks are most effective in missing data imputation. This tutorial aims to provide a brief survey of missing data imputation and introduce the challenges of missing data imputation in different areas such as single-cell gene data analysis time-series analysis, and image processing. The tutorial will also discuss the theory, advantages, and limitations of modern missing data imputation methods and cover low-rank matrix completion, high-rank matrix completion, tensor decomposition, and recent advances of deep neural networks.
