(1) Image recognition

Various statistical models were build to analyze the hand written letters from Large datasets. Bayesian methods and machine learning techniques were applied to perform image recognition. The error rate was improved with the choice of tuning parameters. More introduction about this interesting topic can be found on kaggle.

(2) Modeling and optimization

Mathematical and statistical model development in the case of large scale nonlinear system. Fast algorithms were performed to obtain optimal solutions and estimate parameters. Matlab and C++ packages were constructed and maintained.

(3) Chick embryo gastrulation

Model development and parameter estimation. On the left is the "real world" chick embryo process recorded by a camera. On the right is the computational result. The process is very similar if we consider the motion in a "mirror" way.