AILearners
1.0.0
AIMI-CN Recommended AI learning routes and course notes
We are a group of enthusiasts who love AI learning! Here we learn together, urge each other, and show off together~
We have updated some AI-related notes~ including algorithms, machine learning, deep learning, and natural language processing. More notes will be updated in the future for everyone to learn together~
AIMI-CN AI learning and communication group (there are various AI-related resources) [1015286623] 
Our official account will also push various useful information from time to time waiting for you to follow~
Search the WeChat official account: 'AI-ming3526' or 'Computer Vision' to get more AI direction machine learning resources
Guide you to the beginning of learning machine learning~
Detailed article address
Source code address
Niuke’s online notes on "Swordsman Offer" aim to improve our algorithm capabilities~
Detailed article address
csdn address
Open the door to machine learning for you~
Detailed notes address
Let you use code to experience machine learning ~
Course detailed notes address
Practical Machine Learning Books
Source code and dataset download
Detailed introduction to what are neural networks, CNN, RNN, and GAN~
Detailed notes address
Let's learn the most cutting-edge NLP knowledge together~
CS224n course detailed notes
CS224n learning materials Extraction code: e234
Youtube video link can be found in China on Bilibili
Learn NLP in actual code ~
Detailed notes address
Code and book download address
Prerequisites for viewing:
Choice, method, persistence <br> We all know that there are many resources now. We first choose a piece of information that is truly suitable for us, and then learn with a method that suits us~The most important thing is persistence! ! !
Reprinting a very powerful AI learning route compiled by several very awesome organizations | Complete AI learning route, the most detailed resource sorting!
Learning machine learning requires a certain mathematical foundation, but it is just a little mathematical foundation. Don’t be frightened by these. You are all big shots, just pick up the keyboard.
I will analyze how to learn based on my little experience---
The first thing you need is two giving up:
That’s right, it’s just giving up a lot of information! When we want to get started with machine learning, we often collect a lot of information, such as xx school machine learning internal resources, machine learning from introductory to advanced 100 G resources, xx artificial intelligence tutorials, etc. Many times, we take more than ten or hundreds of G of learning resources, and then put them in a certain cloud disk to store them, waiting to learn slowly in the future. Little do people know that 90% of people just collect information and save information, and have forgotten to open learning after leaving it in the cloud disk for a year or two. The information lying on the cloud disk is often just the self-comfort and "self-safe" sense of security that most people "study hard in the future". Moreover, when faced with a large amount of learning materials, it is easy to fall into a state of confusion. The most direct feeling is: Oh my God, there are so many things to learn! Oh my God, there are so many things I haven't learned! Simply put, the more choices you make, the easier it is to fall into a dilemma of having no choice.
So, the first step is to give up massive amounts of information! Instead, choose a piece of information that really suits you and study it carefully!
Speaking of getting started, many people will think that they should start with the most basic knowledge! Machine learning is a complex technology that integrates probability theory, linear algebra, convex optimization, computer, neuroscience and other aspects. There is a lot of theoretical knowledge required to learn machine learning well. Some people may not have a very solid foundation, so they just want to start with the lowest level of knowledge, such as probability theory, linear algebra, machine learning convex optimization formula derivation, etc. However, the disadvantage of doing this is that it is time-consuming and can easily cause "slack learning" and dispel the enthusiasm for learning. Because judging books and derivation formulas is relatively boring, it is far less likely to stimulate your enthusiasm for learning than building a simple regression model yourself. Of course, it’s not that you don’t need to study basic knowledge, basic theoretical knowledge is very important! It is just that when getting started, it is best to have a systematic understanding of the top-level framework first, and then from practice to theory, check for missing and patch the machine learning knowledge points in a targeted manner. From macro to micro, from overall to detail, it is more conducive to a quick start to machine learning! Moreover, in terms of enthusiasm for learning, it also plays a role of "positive feedback".
Okay, after talking about the two "giving up" before getting started with machine learning, we will introduce the entry route.
I personally think that the mathematical foundations that are needed first: probability theory, matrix theory and calculus. It doesn’t matter if you don’t have it, just learn while watching. Just check if you don’t know what to do.
【Free】Mathematics Teaching Video-Introduction to Khan Academy
| Probability | statistics | Linear algebra |
|---|---|---|
| Khan Academy (Probability) | Khan Academy (Statistics) | Khan Academy (Linear Algebra) |
【Free】Machine/Deep Learning Video- Ng
| Machine Learning | Deep Learning |
|---|---|
| Ng machine learning | Neural Networks and Deep Learning |
Then I recommend a group of domestic bigwigs who are more basic and can record machine learning videos with a little easier to understand than Mr. Ng's.
Machine Learning Practical Practice-ApacheCN Chinese Open Source Organization
The general content is to learn the book "Practical Machine Learning"
Practical Machine Learning Books
Machine Learning Practical Videos
Basically, completing the above courses is considered an introductory one. Next, you can target your interests and directions. For example, you can continue to study the Stanford CS231n course:
CS231n: Convolutional Neural Networks for Visual Recognition
If you focus on NLP, you can learn Stanford CS224n course:
CS224n: Natural Language Processing with Deep Learning
Of course, NTU Lee Hongyi's courses are also very good:
Hung-yi Lee
Of course, there will be corresponding translations of videos in these domestic bigwigs (B site). If you are interested, you will find them yourself.
There are many books on the market that introduce natural language processing technology, and there are also many learning courses and websites on the Internet. However, after investigation, it was found that Stanford's CS224n: natural language processing for deep learning has been favored by the majority of NLP enthusiasts. However, as far as we know, there is no Chinese study notes on the latest CS224n course in 2019. Therefore, in order to better get started with NLP scientific research, we are here to share our learning experience with you, and hope to learn with you.
Natural language processing (NLP) is one of the most important technologies in the information age and a key part of artificial intelligence. NLP applications are everywhere because people communicate almost in language: web search, advertising, email, customer service, language translation, medical reports, etc. In recent years, deep learning methods have achieved very high performance in many different NLP tasks, using a single end-to-end neural model without the need for traditional, task-specific feature engineering. There are two main differences in the 2019 courses compared to the past. First, use PyTorch instead of TensorFlow, and second, the course arrangement is closer. Through this course, everyone will learn, implement and understand the skills they need to do their own neural network models.
1. Understand the basic usage of python
2. Understand basic calculus, linear algebra and probability statistics
3. Have a certain understanding of machine learning
4. Have a strong interest in NLP learning
However, we don’t need to start learning from scratch, which will reduce our interest in learning. Therefore, as long as we continue to make up for the shortcomings of our own prerequisites in the process of learning, we will definitely enter the door of NLP learning.
Organization logo:
