Algorithm learning route, mainly divided into four parts: mathematical basis, programming ability, algorithm basis, actual combat.
Wechat public account search [JackCui-ai] to follow this programmer who loves to post technical dry goods. In this paper, making github.com/Jack-Cheris… Has been included, a line of large factory interview complete test sites, information and my series of articles.
One, foreword
Since writing the article, I have been asked the most question is “algorithm learning route”.
Today, it’s here.
I will take you to see what we need to learn, using this holiday, I even collected the supporting videos and materials, warm male stone hammer ah, if this article is useful, don’t forget three lian oh!
Second, learning route
It is mainly divided into four parts: mathematics foundation, programming ability, algorithm foundation and actual combat.
1. Basic mathematics
In machine learning algorithms, there are two most important basic mathematical knowledge: linear algebra and probability theory.
These two are also university compulsory courses, if the knowledge has already returned to the teacher, it does not matter, where will not learn to fill where.
Linear algebra studies the properties of linear space. Data is usually represented as points in Euclidean space, and these points are mapped to another space after a series of transformations, in which the laws hidden in the data are revealed.
Therefore, linear algebra, as a science of space research, is one of the most important foundations of machine learning.
Video: I recommend a linear algebra lecture by Gilbert Strang, an old professor of MIT.
Did not learn the line generation of students will find this course, speak clearly and intuitive, in-depth line generation of the essence, is not the kind of theory piling up, and the old professor humor, very charming personality.
English is not good, don’t worry, subtitles are in Chinese.
Download address (extraction code: Jack) :
Link: pan.baidu.com/s/1WktC95HL…
** Textbook: ** Recommended linear Algebra and Its Applications by David C. Lay.
For those who like reading, this book must not be missed.
This book introduces linear algebra in detail in geometry, computer graphics, economics, probability theory, signals and systems, differential equations and other fields, giving people an intuitive understanding.
Download address (extraction code: Jack) :
Link: pan.baidu.com/s/10FtcG4mw…
Probability theory is the science of uncertainty. Life is all about probability. Machine learning algorithms need to model real-world situations, and of course use probability theory as a tool.
It’s not a difficult course to get started on, so you can pick a book that’s moderately graded and start.
Here recommend Chen Xiru “probability theory and mathematical statistics”, very detailed, just like listening to an old man recall their own probability and statistics experience, I believe readers will also gain.
Download address (extraction code: Jack) :
Link: pan.baidu.com/s/1P_jIbbW6…
If you have enough time, it is recommended to learn the basics of math first. If you don’t have enough time, skip the basics of math, because when you’re learning an algorithm, you can’t fill in the gaps.
2. Programming ability
Programming languages, at least two, Python and C++ are required.
In my work, Python is mainly used for data processing, algorithm research and model training, while C++ is responsible for project implementation.
The algorithm engineer needs to optimize the reliability and real-time performance of the algorithm for the landing scene, and the C++ engineering ability is essential.
Python is a very friendly programming language that is easy to get started with and powerful enough to be used extensively in the development of machine learning algorithms.
Video: The first video I watched at school was the Little Turtle Python course. It was humorous and involved the basic Python syntax, network crawler, Pygame plane war and other content. It was very rich.
Download address (extraction code: Jack) :
Link: pan.baidu.com/s/1-WasSZey…
Practice is the best teacher. When learning Python, you can find a direction you like and practice.
Real combat, while learning. Such as writing crawlers, doing small games, playing all kinds of interesting algorithms.
Among them, crawler is the simplest, it is easy to have a sense of achievement, let you stick to learning, download novels, download comics, download music, download movies, grab tickets and so on small procedures.
This recommended my crawler project for a Github star with a volume of 11.4K + and fork 4.5K +.
Each actual combat, there are corresponding article tutorials, open source code.
Project address: github.com/Jack-Cheris…
Textbook: Fluent Python is a tricky book to read, but it’s not a handy manual to use. If you’ve forgotten some of the syntax, you’ll learn something.
Download address (extraction code: Jack) :
Link: pan.baidu.com/s/1-YBEOYY4…
C++ is an object-oriented programming language, whether you do algorithms, or development, or testing. C++ is a basic programming language. Once you learn this programming language, you’ll be much more comfortable with others.
Video: This section recommends the free moOCs tutorial. It is a good introduction to the video. The teacher gives vivid, vivid and easy to understand.
It is divided into 7 chapters, each of which lasts 2-3 hours and can be easily finished in half a month. The learning sequence is as follows:
- C++ expedition set sail
- C++ expedition departure
- C++ expedition package part 1
- C++ expedition package (part 2)
- C++ expedition inheritance
- C++ expedition polymorphism
- C++ expedition template
Video viewing address (extraction code: Jack) :
Link: pan.baidu.com/s/19-DHvrNf…
Textbook: C++ Primer, the bible, is a great book.
Download address (extraction code: Jack) :
Link: pan.baidu.com/s/1uyW6kg6J…
Programming language basics, the next is data structures and algorithms.
Data structure and algorithm is programmer’s internal work, every engineer’s required course.
The method of data structure learning, I recommend is to read a book directly, while learning while brushing, at the same time, so that the speed of learning.
Like its title, this is a novel-like introduction to algorithms that is very easy to read and highly recommended.
Download address (extraction code: Jack) :
Link: pan.baidu.com/s/1jQYbWiHM…
Can study together with this book, is “sword finger Offer”, which explains 66 + common data structure questions, analysis ideas, simple and easy to understand.
Download address (extraction code: Jack) :
Link: pan.baidu.com/s/1rNBSsx_-…
Read two books together, easy to get started with data structures and algorithms.
However, the code explained in “Sword Finger Offer” is C/C++, there is no Python version. If you want to see the Python version, you can see the tutorial I organized. Both C++ and Python have implementation and explanation, and the topic has been divided according to type.
Project address: github.com/Jack-Cheris…
Both books are read, the problem is also finished, that is the beginning.
If you want to go further, go directly to LeetCode.
We can start with HOT 100 or Selected Algorithm 200 questions, which are more difficult than “Offer of Sword Finger”, but each question has its own way of thinking and answer.
Insist on brush over 200, most of the interview easily done, completely enough.
B: Yes, it is. C: Yes, it is.
Address: leetcode-cn.com/
Don’t doubt yourself. Brush and practice.
3. Algorithm basis
Congratulations, now that you’ve done all the groundwork, you’re ready to get started on machine learning algorithms.
Machine learning:
Video: Recommended machine learning video by Ng, who is one of the giants in the whole field and has a high academic standing. At the same time, his video is also very friendly to beginners, the first choice.
Download address (extraction code: Jack) :
Link: pan.baidu.com/s/1OglLhzB5…
Teaching material: still be that sentence, light see not practice is not line of. Machine learning actual Combat, theory combined with actual combat, suitable for beginners.
Download address (extraction code: Jack) :
Link: pan.baidu.com/s/1lEz8POdx…
Machine learning in action is implemented using Python2, some details are not detailed enough, I have improved this, using Python3 again, combined with sklearn and more fun examples, to explain.
Total web reading: 500W + :
Corresponding Github open source code Star 3.3K +, fork 3.1K +.
Read online:
cuijiahua.com/blog/ml/
I also packaged the series as a local PDF, which I like to view offline or download directly.
Download address (extraction code: Jack) :
Link: pan.baidu.com/s/11OI0NZ_F…
Deep learning:
Deep learning is a subfield that today’s algorithm engineers can’t get around, a subset of machine learning.
Video: I still recommend the deep learning video by Ng enda, which is also very friendly to beginners.
Download address (extraction code: Jack) :
Link: pan.baidu.com/s/1TShDS2_j…
Teaching materials: To be honest, DEEP learning, I have not read the book, is the video + Github open source project to learn, but known as the bible of deep learning “flower book”, can prepare a book.
Download address (extraction code: Jack) :
Link: pan.baidu.com/s/1drDJUf9O…
Deep learning framework:
There are many deep learning frameworks, such as Tensorflow, Pytorch, Paddle, MXNet, Caffe, etc.
In my work, I used Pytorch the most, followed by Tensorflow.
Pytorch is recommended for beginners to learn Pytorch first. You can directly watch Yunjey Choi’s Github tutorial to get started.
Project Address:
Github.com/yunjey/pyto…
Pytorch Deep Learning framework learning, you can also see my Pytorch deep learning practice series tutorials, garbage classification, image segmentation and other small projects combined with actual practice.
Github open source code Star 400+, fork 250+
Project Address:
Github.com/Jack-Cheris…
I also packaged the series as a local PDF, which I like to view offline or download directly.
Download address (extraction code: Jack) :
Link: pan.baidu.com/s/1PXkcKJa-…
4, in actual combat
Actual combat, the article repeatedly mentioned so many times, only these are far from enough.
Because most of the time, you are following the ideas of the video or the article, which lacks the independent thinking process.
After learning so much, you haven’t finished a project independently, nor have you thought independently about how to deal with data, how to analyze problems and what algorithm to use to solve problems.
The job of algorithm engineer is also highly competitive. In order to stand out from the crowd, we need to complete some projects independently as a team or individual. Only in this way can we be more competitive.
For the Student Party, the easiest and most direct way to do a project is to enter a competition.
Recommend two places to participate in the competition, one is foreign Kaggle, the other is domestic Ali Yuntian Pool.
The prize money of the two competitions is also very rich, according to their own preferences, you can participate.
Kaggle：www.kaggle.com/
Ali tianchi: tianchi.aliyun.com/competition…
You get to play and win money. Why not?
Three, data packaging
I have packed all the materials mentioned in this article and can take them away directly.
Download address (extraction code: Jack) :
Link: pan.baidu.com/s/12tbVrUF0…
PS: Have the ability to support the legal version, the data feel good, you can buy a wave of legal support.
Four, omg
This article is also my study route.
How long you need to learn depends on your motivation.
If you want to go to a good company, but the current hard power is not good, I think it is necessary to try, the level of technical ability determines how far you can go, the level of platform determines how high you can fly.
If you can enter your favorite company through your own efforts, you must not slack off. Career growth is the same as learning new technology, if you do not advance, you will fall back.
You will often find that the stronger the people around you, the harder they work. The highest form of self-discipline is to enjoy being alone.
It’s not easy to create. In this hardcore issue, your three lines are the biggest motivation for Jack Cui’s creation. See you next time!
This article will be updated continuously. You can find it on our wechat official account by searching [JackCui-ai]. GitHub github.com/Jack-Cheris… Welcome Star.