Weichung Wang

Dr. Weichung Wang is a professor at the Institute of Applied Mathematical Sciences, Department of Mathematics and Data Science Degree Program of National Taiwan University. His research interests include medical artificial intelligence, matrix computing, and scientific computing. He received the Nian-Tsz Award, NTU Outstanding Professor Award, MOST Outstanding Young Scholar Project, MOST Futuretech Award, and an Honorary Research Fellow of the National Center for Theoretical Sciences. Professor Wang serves as the president of the Taiwan Society for Industrial and Applied Mathematics (TWSIAM) and has been the secretary of EASIAM and TMS. He also serves on the editorial board of international journals, including the SIAM Journal on Mathematics of Data Science, and is actively involved in major international conference organizations.

Ph.D. in Applied Mathematics, 1996
University of Maryland, College Park, USA

B.S. in Applied Mathematics, 1989
National Chiao-Tung University, Taiwan

Research Summary

I am interested in developing numerical algorithms and software for medical AI, data sciences and scientific computing. My researches usually involve artificial intelligence, numerical linear algebra, computational optimization, parallel computing, and their applications.

Medical AI

Professor Weichung Wang founded MeDA Lab (Medical Data Analytics Laboratory, http://meda.ai) to develop medical AI Engines and medical AI Workflows. The former extracts hidden information from high-dimensional medical image and numerical datasets. The latter turns the information into clinical intelligence. MeDA Lab works with physicians and industry partners worldwide to boost intelligent and precision medical software solutions.

Matrix Computations in Computational and Data Sciences

We consider large-scale linear systems, nonlinear eigenvalue problems, and matrix factorizations arising in numerical simulations and data analytics. Applications include 3D photonic devices, big data analysis, and healthcare. The main focuses include Krylov type algorithms, randomized methods, and accelerations on parallel computers with GPU.

Data-driven Modeling and Statistical Computing

Many computer experiments conduct performance analysis and optimization while only objective function values are available. The main research focuses are the design of computer experiments and surrogate models-assisted techniques with an emphasis on software auto-tuning. We also develop efficient methods for statistical computing and optimal experiment designs.

GPU and High-performance Computing

We study how GPU, CPU, and heterogeneous CPU-GPU clusters can be used to accelerate scientific computations. The focuses include CPU-GPU accelerated solvers for linear systems and eigenvalue problems, fast medical image reconstructions on computed tomography, and particle swarm optimization with applications in medical and statistical sciences.

Numerical Methods

Computer Aided Learning