Through my contact and understanding for more than half a year, combined with the needs of some bigwigs, I put forward my own views on how to get started with machine learning.
If you want to be a machine learning engineer, algorithm engineer and data mining engineer in the future. Take a good look at some of my ideas!
Learn machine learning first, then learn deep learning (suggestion from the great god)
Proficient in at least one direction of deep learning: CV and NLP . During the period, you should review the data structure, mathematical foundation and strengthen your programming ability by joking around.
Machine learning integrates complex technologies such as probability theory, linear algebra, convex optimization, computer, neuroscience, etc., but in fact, the most useful knowledge of linear algebra and advanced mathematics is the knowledge of these two. The foundation of the subject is firmly established.
1. Mathematical skills: probability theory, linear algebra, advanced numbers, information theory (mainly the part of information entropy).
2. Data structure: tree, stack, linked list, queue, graph! Take your time. (There must be awareness of optimizing the complexity of the algorithm)
3. Programming ability: Please transfer to the camp of leetcode and race code to hone it.
Books: "Statistical Learning Methods" (Li Hang), "Machine Learning" Western Book, "Deep Learning" (Yoshua+Bengio+&+Ian+GoodFellow), "PRML" is not very friendly to Xiaobai, "Data Analysis with Python"
When I was studying the watermelon book, my head would hurt after a whole day of learning. The formulas were obscure and difficult to understand, so it was not suitable for novices to learn, not directly learning the watermelon book.
I highly recommend ✨✨Andrew Ng's introductory course on machine learning "Machine Learning" on Coursera. You may find this in station B, but remember to do it with after-school exercises. It is not enough to learn the theory without programming. the ~
The courses taught by Professor Li Hongyi from National Taiwan University are also good. The lectures are lively and interesting, and some of the content can complement each other with Ng. Tang Yudi’s courses are also suitable for beginners. They are relatively easy to understand. If you have time, it is recommended to watch them together.
Deep learning tools tensorflow and pytorch, you can find videos to learn, you can buy books to read, just learn one of the frameworks, and learn to read official documents.
You can improve your actual project experience by playing competitions: Jingdong, Tencent, Tianchi, kaggle, etc.
Finally, if necessary, please pay attention to the public H: AI Technology Planet, reply 211
Included in the acquisition: deep learning neural network + CV computer vision learning (two major frameworks pytorch/tensorflow + source code courseware notes) + NLP, etc. Applicable
people
① students who are preparing for graduation thesis
② AI algorithm engineers who are ready to change jobs and are looking for a job
③ self-study and People who are going to switch to the AI field
④Those who want to consolidate the core knowledge of AI and fill in gaps
Thesis/study/employment/competition guidance + Daniel technical answers