标签：
To start this post, it is the best to first let me tell you a story that had happened in the year 2013. It was in the summer of 2013, when I was near finish the study of PhD, preparing the final seminar and the submission of my thesis for external review. I got an opportunity of an interview with a professor. At that time, I did not have a journal paper yet, as my first journal paper was still under review by Pattern Recognition Letters. Then the professor raised a question in the interview: "I went through your CV, and it seemed to me that your last peer-reviewed publication was a paper in AVSS 2012 conference, which is more than a year ago. I wonder why you do not have any publications for a year. Does it mean you have been lazier now?" "No." I said, "In fact, I work much harder in this year. After I publishing a few conference papers, I felt that I can easily have a paper in conferences such as AVSS. However, if I spend too much time on writing papers, the time for literature review is not enough. If I did not read enough books to obtain more knowledge and spend all time on writing papers, my level will remain to be the same despite the number of publications increases. Once I finish my degree and no longer be a student, I am afraid I will not have that much free time to read textbooks as thick as brisks and then I will eventually encounter some bottlenecks for further research progress… My time mostly nowadays is spent on reading textbooks for measure theory and functional analysis, and I feel that my level has been increasing… we have a paper under review by Pattern Recognition Letters and I believe the quality of the paper exceeds all my previous publications… " However, the professor did not appreciate this response, and eventually he said I would not be able to get a position in his team.
I am not intending to discuss for whether I think that style of doing PhD is good or not, as this matter is always subjective. However, as I expected in 2013, I would probably do not have as much time for book reading as in 2013 when I was still a student. Life is not always that easy! We are all busy with a number of commitments. However, to be able to further progress to a higher level, I think I should spend some time at night for book reading. Thus recently I started to read a very classical book, Statistical Learning Theory, written by Vladimir Vapnik, the inventor of support vector machines. Before reading this book, I have read his book Estimation of Dependences Based on Empirical Data. This is one of the best book I have read so far. The book Estimation of Dependences Based on Empirical Data is written in a very special style. Besides telling you the knowledge and techniques, Vapnik paid much attention to draw the overall picture of the discipline, the development of pattern recognition since 1930s when Fisher developed the first recognition technique, Fisher‘s linear discriminant analysis, as well as the philosophy behind the ideas. I highly appreciate him to spend more than 30 years to develop the SVM theory. After reading some part in Statistical Learning Theory, I can feel that the author wrote the book in a very rigorous and careful style. However, some terminologies seem a little different from other books, causing some difficulties to understand. I will keep on sharing with you about my understanding of the books in future posts.
?
?
?
?
Reading “Statistical Learning Theory”
标签：
原文地址：http://www.cnblogs.com/jingxinxu/p/5370965.html
