i
i
i
i
i
i
i
i
Acknowledgments
I would like to first thank Prof. Jinpei Xu at the Harbin Institute of Technology (HIT)
and Prof. Wen Gao at Peking University, although they did not directly contribute
to the research presented in this book. Prof. Xu was my supervisor when I was a
master’s student at HIT. My first research project, continuous speech recognition us-
ing neural networks, was done with Prof. Xu. Every week we spent several hours
discussing all the details, including the initial idea, the implementation, the experi-
mental results, and planning for the next step. I really appreciate what I learned from
Prof. Xu about the basic skills for doing research.
Prof. Gao was my Ph.D. supervisor. When I completed my courses and began
my research, Prof.