Hyekyoung Lee

I received my Ph.D. degree at the Machine Learning Group of Computer Science and Engineering in POSTECH , Korea. My Ph.D. advisor was Prof. Seungjin Choi. The title of my thesis is "EEG classification with nonnegative matrix factorizations," which contains several modified version of NMF such as nonnegative tensor factorization (NTF), kernel NMF, group NMF and semi-supervised NMF. At that time, I was interested in the EEG-based brain computer interface (BCI) system, especially the feature extraction methods from noisy EEG data. So, I modified NMF for the EEG data.

I'm interested in brain connectivity. Changing my research area, my first question was how to visualize complex brain network effectively and help my colleagues who are psychologists and medical doctors interpret it easily. To reduce its complexity, the thresholding is unavoidable. But the problem is that there is no widely accepted criterion on it. Fortunately, we could find the way without thresholding from the persistent homology, thanks to Prof. Moo K. Chung. Now, I, Prof. Hyejin Kang and Prof. Dong Soo Lee are developing new methods based on persistent homology by extending our previous one to multi-modal imaging analysis.