Skip to content

Featured Research

Professor Shih-Feng Huang

  • Time Series Analysis
  • Financial Engineering
  • Financial Econometrics
  • Risk Management
  • Data Science

Professor Tsung-Shan Tsou

  • Major cooperative partners include the Institute of System Biology and Bioinformatics of National Central University, Taipei Cathay General Hospital, Chungli Landseed Hospital, and Taipei Medical University. Dr. Tsou also actively participates in the execution of joint research programs between National Central University and Landseed Hospital, and uses data from cohort research to explore the related factors of occurrence and disappearance of metabolism disorder syndrome together with doctors of Landseed Hospital.

Professor Chun-Shu Chen

The primary research focus lies in the development of spatial statistical theory and methodology, encompassing the following four key directions:
  • Development and application of spatial model selection criteria
    • This research aims to establish objective criteria for fairly comparing various spatial prediction methods, facilitating the selection of the most appropriate model for specific datasets. These criteria can be further extended to practical selection problems in areas such as environmental assessment and resource planning.
  • Estimation bias and statistical inference in spatial regression models
    • This line of work investigates the estimation bias and inferential inaccuracies that may arise when explanatory variables are correlated with spatial random effects. Semiparametric methods are employed to improve the accuracy of coefficient estimation and statistical inference.
  • Efficient analysis strategies for high-dimensional spatial data
    • To address the computational challenges posed by large-scale spatial domains and high-dimensional data, spatial partitioning techniques are utilized to construct scalable models. These approaches approximate underlying spatial processes while reducing the computational burden of large matrix operations, thereby enhancing both analytical efficiency and predictive accuracy.
  • Spatial modeling and risk assessment of extreme events
    • Focusing on environmental extremes such as heavy rainfall and extreme temperatures, this research integrates spatial statistical models with extreme value theory to analyze the spatial distribution and mechanisms of such events. The resulting models support risk assessment, policy formulation, and early warning systems for extreme weather and climate-related hazards.

 

Professor Li-Hsien Sun

Professor Shao-Hsuan Wang

  • Dr. Wang’s primary research interests include high-dimensional data analysis, dimension reduction and model selection, image processing, and survival analysis. With the rapid advancement of computational technology in the era of big data, classical statistical methods are often inadequate for tackling emerging challenges. One of Dr. Wang’s major research efforts focuses on developing new statistical inference methods tailored to the complexities of large-scale data. For instance, he investigates how cross-data matrix techniques can relax the stringent consistency conditions required for principal component analysis (PCA). He also studies variable selection in generalized linear models using three-parameter Bayesian models, with applications spanning bioinformatics, medical imaging, and survival analysis.
  • Image processing represents another key area of Dr. Wang’s research. His work addresses issues such as model bias, data clustering, and data visualization. In high-volume datasets, noise becomes a significant factor that cannot be ignored. In many applied settings, noise may create misleading signals and result in erroneous scientific interpretations. For example, in 3D reconstruction of molecular structures, high levels of noise can mislead machine learning algorithms into producing spurious molecular models. To address this, Dr. Wang has developed methods based on extreme value theory to distinguish true signal from noise. His techniques enable effective clustering and 3D reconstruction even in highly noisy environments. In the domain of data visualization, Dr. Wang has contributed a two-stage manifold learning approach that improves upon current techniques for visualizing complex data structures.