In 2017, artificial intelligence has been included in the State Council's artificial intelligence development plan. Artificial intelligence will become a new science and technology force that will promote the development of China, and it will play an increasingly important role in the future. For those who want to engage in the AI ​​industry, For example, how to quickly and in-depth knowledge of machine learning is particularly important.
Below we have arranged the basic flow for the next machine learning from entry to advanced, mainly focusing on the advanced part.
First, machine learning essential knowledge
The figure above shows the relationship between artificial intelligence and machine learning, deep learning, and evolution over time. Nowadays in the industrial world, machine learning and in-depth learning all play a decisive role. The problems solved by the two are different: Machine learning is good at solving classified prediction problems of structurable data, such as house price prediction, disease prediction, etc. Learning focuses more on some issues of unstructured data, such as image classification, text classification, and so on.
As an introduction to machine learning, of course, recommended courses and books, here are some basic books and courses that you would like to get started with machine learning.
1, "The beauty of mathematics", author Wu Jun. This book is very suitable for entry, it is very popular, and there is no such thing as a series of formulae. If you read this book well, you will find that the algorithm is very interesting and can solve many problems.
2. “Introducing Data Analysis in a Simple Way†This book mainly explains some basic concepts and knowledge of data analysis, and is also suitable for entry-level reading.
3, "MySQL must know," this book is mainly to explain MySQL, you want to get started AI, basic SQL still have to understand.
4, "machine learning", the author Zhou Zhihua, this book can be said to be a magic weapon for machine learning entry, not only an introductory book, and worthy of your late or even late large-scale continuous study, entry must see, advanced is equally important.
5. Stanford Open Class: Li Feifei - Deep Learning Computer Vision. Li Feifei is Professor of Stanford Computer Department, Stanford Artificial Intelligence Laboratory, and Director of Vision Lab. At the same time, Li Feifei served as chief scientist in Google Cloud. This course is a deepening course in machine learning. It mainly introduces the application of deep learning (especially the convolutional neural network and related frameworks) in the field of computer vision. It covers the specific structure of a variety of neural networks and training application details, as well as focusing on Image recognition, object location, object detection, image style migration, image understanding description, and video content recognition.
6, other courses, you can choose to buy one or two courses on the live platform, follow the teacher from beginning to end, directly listening to the above Li Feifei course may be difficult to keep up with, one or two video tutorials can still speed up learning Speed, the other thing is to mention that the entry does not need too much, but it is best to buy a high quality, so that generally there will be teachers answering questions can also have a lot of people to communicate, will not learn to learn to persist, persist is also very important.
Second, machine learning advanced knowledge
With the basic skills of the above learning, you can consider advanced learning, the above entry knowledge may take you 3 months or even longer, but not too tangled, do not understand the above issues, in progress The section can also continue to study.
Advanced machine learning, the knowledge to be learned will begin to be biased towards specific algorithms and programming practices. This part of the proposal is not too much, you can take each knowledge point one by one breakthrough method, each algorithm one by one to break the way, give everyone a few recommendations Reliable books:
1, "machine learning", author Zhou Zhihua. Yes, it is, don't underestimate this book, he will accompany you throughout the entire learning process, this book personally feels that not every algorithm is speaking well, for algorithm breaks all, it is recommended like a decision tree, random forest, Naïve Bayes and the integration algorithm take a good look. This part is well written, like a decision tree, basically according to what it says, and it can be implemented by hand. It is very good.
2, "statistical learning methods," author Li Hang. This book can be said to be an advanced weapon. If you want to understand the algorithm in machine learning, this book must be read well. It is suggested that the SVM and KNN algorithms in this book should be taken a good look. This book is very Clear understanding, many books like SVM omit many steps, not enough thoroughness.
3, "Machine Learning Actual Combat", carrying the book carrying the girl with a basket, this book is very suitable for combining the above two books together to see, personally think that it is machine learning "Three Musketeers", these three books look good Look, follow the steps to learn more about the code.
4, "using python for data analysis", this book is very important, above is the principle of the algorithm, if you really apply the algorithm to practice, there is a considerable part of the workload is to process and analyze the data, machine learning Most of the data in the data can be converted into DataFrame data for modeling. Data analysis has learned that practical applications will not be far behind. However, with respect to data analysis, practice is very important. Reading books alone is not enough. This section suggests finding a solution. It will be better to follow the lesson or buy a lesson.
5, "Python natural language processing", this book is mainly about natural language processing, but also a more important branch, there are interested can learn
6. "Neural Networks and Deep Learning" by Michael Nielsen. This book can serve as an introductory and advanced level of deep learning. This book should take a good look, speak very well, and teach you deep learning.
7, other courses, this part of the need to learn more than a lot of entry, to learn this part of the need to spend a lot of time to read, but if you have been reading, there is no practical project, it will be difficult to stick to it, the proposed small Partners are looking for a course that you will learn from beginning to end. You will learn while reading and you will have better results.
Third, machine learning learning methods
The above has listed some of the necessary knowledge and courses for learning machine learning that Xiaobai has or has basic skills. With these as reserves, your hardware resources are enough, which is equivalent to teaching materials and courseware, but learning With these things is not enough, how to learn, how to learn efficiently and the steps and focus is also very important, the following will give you a list of some issues in the learning process.
1. Do not fall into Shushan and combine practical theory with important
After enumerating the above studies, many people may head into Shushan and study hard. After reading a book and then finally discovering something, they all understand something and do not understand it. This efficiency is very low. If you are Engaging in academic work is not within this scope. I believe most people are still partial applications and engage in related work. In the IT industry, practice can learn something. Don't read books all the time. You can try to write code while you look and realize a small formula and small algorithm. , This progress is faster.
2. Adopt parallel learning instead of serial
This map can be used as the basic flow of learning, but it is not necessarily necessary to have it in front of it. You can learn the basics while writing the code. You can also do the competition. This may be more painful in the early stages, but you will learn more quickly.
3, establish a knowledge framework to repair knowledge loopholes
The above picture is sklearn's machine learning algorithm notes, you can create similar notes in the study, help to establish the entire learning framework, do not know how to, and continue to learn from the details.
4. Find a course that suits you and follow it from beginning to end
If you are an office worker or if you haven’t had a teacher to teach you this class, it is very important to report a course that suits you from beginning to end. Don’t be too confident in your self-control, nobody and you. Learning together, you haven’t seen any progress since you’ve studied for a long time, and you’ve lost interest in it. So it’s also important to find courses that can be taught to you from beginning to end. you.
We are manufacturer of Micro USB 3.0 Male in China, if you want to buy Micro USB 3.0 Straddle Type,B Male Straddle Type 0.8mm,MICRO USB 3.0 B MALE please contact us.
USB3.0 is a USB specification, which was initiated by Intel and other companies. The maximum transmission bandwidth of USB3.0 is as high as 5.0gbps (500MB / s).
While maintaining compatibility with USB2.0, USB3.0 also provides enhancements: significantly increased bandwidth (up to 5Gbps full duplex); better power management;
more power; faster device identification; and higher data processing efficiency.
The reason why USB 3.0 has the performance of "speeding" is entirely due to the improvement of technology.
Compared with USB 2.0 interface, USB 3.0 adds more physical buses in parallel mode.
You can pick up a USB Cable and look at the interface.
On the basis of the original 4-wire structure (power supply, ground wire, 2 pieces of data), USB 3.0 adds 4 lines for receiving and transmitting signals.
So there are eight lines in the cable and on the interface.
It is the additional 4 (2 pairs) of lines that provide the bandwidth required for "superspeed USB" to achieve "over speed".
Obviously, two (1 pair) lines on USB 2.0 are not enough.
In addition, in the signal transmission method, the host control mode is still used, but the asynchronous transmission is changed.
USB 3.0 makes use of two-way data transmission mode instead of half duplex mode in USB 2.0 era. In short, data only needs to flow in one direction, which simplifies the time consumption caused by waiting.
In fact, USB 3.0 does not take any rarely heard of advanced technology, but theoretically increases the bandwidth by 10 times. As a result, it is more friendly and friendly. Once superspeed USB products come out, more people can easily accept and make better customized products.
Micro USB 3.0 Male,Micro USB 3.0 Straddle Type,B Male Straddle Type 0.8mm,MICRO USB 3.0 B MALE
ShenZhen Antenk Electronics Co,Ltd , https://www.antenkwire.com