Press Releases

HKBU mathematician gives a plenary talk on using Deep Neural Networks for Image Processing

2020.10.6

Prof Xuecheng Tai from the Department of Mathematics, Hong Kong Baptist University, was invited to give a plenary lecture at the ALGORITMY 2020 Conference held on 10-15 September 2020. ALGORITMY (Algorithms) is the oldest Central European series of international high-level scientific meetings devoted to applied mathematics and numerical methods in computational sciences and engineering.

At the Conference, Prof Tai presented his team’s recent research on using Deep Neural Networks for Image Processing. Prof Tai’s team used neural networks for semantic image segmentation to find interesting objects and shapes within certain given images. Through this research the team invented a new technique that guarantees the shapes found by their neural networks must be convex and star shapes if the true objects are convex and star shapes. This is important for real world applications, as many real objects are convex or star shapes, yet existing networks fail to guarantee this. The results are based on the following research projects: 

  1. Tai, X. C., Initiation Grant for Faculty Niche Research Areas (IG-FNRA) (RC-FNRA-IG/19-20/SCI/01): Convolutional Neural Networks for Medical Image and Data Analysis.
  2. Tai, X.C., NSFC/RGC Joint Research Scheme (N_HKBU214/19): Mathematical modelling and analysis of deep neural networks for solving structured differential and integral models.

Please click here for more details of the ALGORITMY Conference.