Combined with their own learning experience, summarize how to learn machine learning. In fact, my own learning process is very chaotic and painful. I don't know if I can get into it. I hope it can help others who want to teach themselves machine learning but can not find the direction.
1、 You can read some popular and comprehensive things.
The popular science article "machine learning and data mining" written by Professor Zhou Zhihua of Nanjing University is not bad, and the difference between machine learning and data mining is very good. In addition, the history and Prospect of machine learning are explained. At the end of the paper, some important conferences and journals are also given.
The beauty of mathematics written by Wu Jun (also very good at the top of the tide) may be a comprehensive science popularization of search, natural language processing and machine learning, but machine learning has a lot to do with these fields, so it is also a good introduction science popularization for those who study machine learning.
2、 You can know the best people in the field. Professor Yang Qiang, who once listened to Hong Kong University of science and technology, said in a lecture that a good way to interview a student is to ask him who are the best people in this field and what is the representative work of each of them, so as to test whether a person really likes and pays attention to this direction. I think it makes sense. There are some people on the Internet who have summed up some great people. Of course, these summaries are all personal opinions. Just take a look at them. (Note: the link below is not necessarily the original source)
Http://blog.sina.com.cn/s/blog f7cc3a0100qktd.html (foreign)
Http://blog.sina.com.cn/s/blog a6b58ce01017jy3.html (domestic)
Http://blog.csdn.net/yihaizhiyan/article/details/6795073 (there are some in machine vision)
3、 You can learn about the important meetings in the following fields. As mentioned in the previous week's article, there are also various versions of analysis on the Internet. There are two columns
Http://blog.csdn.net/blow_jj/article/details/24153005 (for artificial intelligence and machine learning)
Http://taoo.iteye.com/blog/1052495 (for database and data mining)
4、 We can systematically take a machine learning course. Andrew NG's machine learning course of Standford is very good. There is a video of his open class on the Internet. It's like a spring breeze to take his class.
Course homepage:
Http://cs229.stanford.edu/ (there are handouts in it, so it's better to print them out and check the video)
Video with Chinese subtitle translation in Netease open class:
http://v.163.com/special/opencourse/machinelearning.html
This is the version in coursera project. It seems that there are more contents than Netease, at least more parts with recommended algorithm
https://class.coursera.org/ml/lecture/preview/index
5、 Can systematically read a machine learning and data mining teaching materials, system learning is very important. There are many recommended textbooks on the Internet
Pattern recognition and machine learning by Christopher M. Bishop
The elements of statistical learning by Trevor Hastie
Richard O. Duda's pattern classification
Tom M. Mitchell's "machine learning", it's a little old
Data mining: concepts and technologies by Han Jiawei
6、 For some important classic articles, please read them. The following list of materials can be referred to
http://www.newsmth.net/bbsanc.php?path=%2Fgroups%2Fsci.faq%2FNLP%2F1%2FM.1225371502.h0
In the process of learning, it is necessary to have a reference book of statistics at hand. Of course, there is also a sharp tool Google. You can check the concepts you don't understand at any time.
7、 In fact, the most important thing to learn machine learning is to practice, practice to know.
==================
by lcj ,2012 -10 -14