https://www.toutiao.com/a6709094073963119115/
Github
Now that more network resources, especially on Github more generous glory
Or that there are many resources that can be learned on the B station
This article focuses on some of his views from artificial intelligence to learn the overall framework
If omissions, please forgive
Foundation Stage (mathematical basis)
In Python, for example, we must first lay the Python basis , then further understanding of Python
Familiar Numpy scientific computing library, familiar Matplotlib visualization library database and data analysis Panda , overall, that is to learn to master the knowledge of Python
Lay the foundations of mathematics: learning the basic theory of mathematical analysis, advanced algebra calculus and so have to learn, especially for linear algebra matrix theory, probability theory to reinforce learning
Part of understanding computer vision or natural language processing tools, look at the examples of actual combat , because they have an intuitive understanding, to understand some of the game
Entry phase (machine learning)
Machine learning classical principle of derivation learning algorithm , the application of machine learning analysis
Skleran library to learn, to know how to optimize parameters and iterative gradient
Learn to characterize data processing and data according to different
, Characterized in text, graphical features, time series analysis model visualization
Conduct comparative analysis of the different attractions of learning algorithms, understanding how to win the competition, how to win, to understand large corporations to solve project examples
Advanced stages (deep learning)
Neural networks, convolution, recurrent neural network learning
Learning Opencv library, understand the mainstream framework
Based Tensorflow or Caffe conduct combat training, such as learning Keras
Use Opencv do some small projects
Natural language processing and also learn practical
Ultimate Advanced (solid foundation)
We know the origin of all the problems and the advantages and disadvantages
To analyze root out classic paper reproduction
Learn relatively advanced skills and deep learning model
Depth study of the neural network
Understand the future trends
Comprehensive combat, plan for the future and lifelong learning