Advanced Computing and Modeling

Computer and information technology has fundamentally changed our everyday lives through the computing, collection, processing and transmission of vast amounts of data. The power of computation is increasing by leaps and bounds. Researchers @ HKBU SCIENCE are using new ways of creating, processing, managing and modeling this avalanche of data.

Model Checking and Dimension Reduction

To analyze the relationship between one variable and others, the statistical method of regression analysis is often required. However, to establish a correct model, a checking should be accompanied to avoid misleading further analysis, and at the same time the relatively unimportant data should be taken out to facilitate further analysis. “Model checking and dimension reduction in regressions” not only increases the effectiveness of model checking, but it is simpler and faster than other existing regression methods.

This innovative result in the statistics theory will be applied in the areas of biological medicine, genetics, music and business. For instance, if we want to check the factors leading to diabetes, we can use the technology of model checking and regression to extract from some 100,000 genes that affect the patient, for analysis.

Professor Zhu Lixing
Department of Mathematics

In the model checking method, we investigate the use of globally smoothing method that requires no fitting of non-parametric functions. Hence the effectiveness of checking is higher. In respect of dimension reduction, the regression method can be used to solve non-linear problems easily and quickly.   

 

 

     

Large-scale Storage Systems for Next Decade

Large-scale storage systems play a key role in today’s IT infrastructure and the need of data storage is expected to reach an astonishing 4 billion Terabytes in 2020. In view of unbalanced growth of storage capacity and computing capability, a set of newly designed parallel techniques that can fully exploit the computing resources available in all storage nodes to satisfy the computing requirement of the storage system at post-Exascale without affecting other computing task from customers is under development. It could help designers of disturbed storage systems overcome the major computational hurdles, and therefore make secure and efficient post-Exascale storage systems a reality in the near future.

Dr Chu Xiaowen
Department of Computer Science

The ultimate purpose of this project is to harness the power of parallel computing to solve the potential computational bottlenecks in secure large-scale storage systems. recovery from food waste, and reduces the environmental impact.