As a graduate student of computer science, I solved a problem of course scheduling in an e-learning system. In addition, I spent a great deal of time studying non-CS topics, including current learning theories and teaching strategies, and developed a courseware sequencing engine tailored to pedagogical needs. I am very proud of this achievement; more so because my paper was selected as one of the best paper candidates at the National Computational Symposium 2003 (NCS’03). This research also helped me understand how computer science techniques can be applied to other fields.
        After graduating, I chose Bioinformatics as my field of study because it presents so many challenges. As a research assistant of Dr. Wen-Lian Hsu and Dr. Ting-Yi Sung, I have had an opportunity to study protein local structure prediction, and utilize two machine learning methods for local structure prediction, namely, a knowledge-based method and a neural network method. My results in the paper submitted to CSB’05 show that the combination of the two methods outperforms other approaches used in Protein Blocks and the structural alphabet of HMMSTR. This project not only strengthened my interest in machine learning approaches, but also raised a number of questions for me. For example, would the results be better if I were to select other machine learning approaches? Also, would there still be a positive result if I were to apply the same techniques in other fields? Clearly, this is a constant learning process.
        In the future, I will study the following subjects:
        1.Protein structure prediction
        Since I have already had some preliminary success in protein local structure prediction, I plan to extend my work to ab initio protein 3D structure prediction. Though existing research has shown some positive results, a satisfactory level of prediction accuracy cannot be guaranteed. I believe my previous experience with machine learning methods would enable me to improve this accuracy
        NMR (Nuclear Magnetic Resonance) spectrometer, a facility used extensively by chemists and biologists, enables us to study the physical and chemical properties of a protein. Using experimental data of the Nuclear Overhauser Effect (NOE) and Resid-ual Dipolar Coupling (RDC) in a NMR spectrometer could enhance the accuracy of 3D structure prediction. I am also very interested in this aspect of protein structure research.
        2.Machine learning methods
        Because of my CS-background, I am interested in devising and applying machine learning approaches to solve existing problems. Since different approaches have different strengths and weaknesses, it is important to understand their capabilities and use them appropriately. Moreover, developing a hybrid method that takes advantages of several complementary approaches is a challenging task that I would like to undertake. I believe that understanding machine learning approaches thoroughly and using them appropriately would benefit my research enormously.
        Although these topics are my major interests, I am not restricted to them and would rather keep an open mind on other tasks and challenges in the field of Bioinformatics. After obtaining my Ph.D. degree, I will devote myself to academic research in the field of Bioinformatics. This is my long-term goal. I know that dedication and hard work will lead me to a promising future.