What is the best way to learn machine learning

Click on the above “ Xiaobai studies vision ”, Optional plus " Star standard " or “ Roof placement ”

 Heavy dry goods , First time delivery 

Before we start , I want to talk about the relationship between machine learning and data mining .

As shown in the above figure, data mining is only one of the fields involved in machine learning , Machine learning and pattern recognition 、 Computer vision 、 speech recognition 、 Statistical learning and naturallanguageprocessing .

Machine learning is ML, It's an interdisciplinary subject , Probability theory 、 statistical 、 Approximation theory 、 Convex analysis 、 Algorithm complexity theory and other disciplines . To study how computers simulate or implement human learning behavior , To acquire new knowledge or skills , Reorganize the existing knowledge structure to improve its performance .

Machine learning is a relatively young branch of artificial intelligence , Machine learning is also divided into supervised learning and unsupervised learning , At the same time, as artificial intelligence is more and more valued and hot , Deep learning is also a new field of machine learning .

machine learning , Start with the knowledge list

When we learned to drive on the first day, we would not go straight on the road , But you should learn the basic knowledge first , And then conduct the on-board simulation .

Only a comprehensive understanding of knowledge , Only in this way can we ensure that even if we encounter problems in our future work , You can also quickly locate the problem , Then find a way to correspond and solve .

So I made a list of the knowledge for getting started with machine learning , They are the general process of machine learning 、 Ten algorithms 、 The triple realm of algorithm learning , To start our learning journey .

One 、 The basic process of machine learning

Quote the boss's explanation ：

A computer program is said to learn fromexperience E with respect to some task T and some performance measure P,if itsperformance on T,as measured by P,improves with experience E.   —Tom Mitchell

Simply speaking , Machine learning is aimed at real problems , Use the data we input to train the algorithm , The algorithm will generate a model after training , This model is a description of the law of data capture for current problems . Then we import the model further into the data , Or introduce new data sets for evaluation , Adjust the algorithm according to the result , Form feedback and optimization closed loop . The whole process machine is constantly learning 、 Training and optimization iterations , This is also where machine learning is powerful .

Two 、 Ten algorithms of machine learning

For machine learning and data mining tasks , Data scientists have come up with various models , In many data mining models , International authoritative academic organization ICDM（the IEEE International Conference on Data Mining） Ten classic algorithms were selected .

For different purposes , I can divide these algorithms into four categories , So that you can better understand .

Classification algorithm ：C4.5, Naive Bayes （Naive Bayes）,SVM,KNN,Adaboost,CART

clustering algorithm ：K-Means,EM

Correlation analysis ：Apriori

Connection analysis ：PageRank

1. C4.5

C4.5 The algorithm is the one that gets the highest number of votes , It can be said to be the first of the ten algorithms .C4.5 Is the algorithm of decision tree , It creatively prunes the decision tree in the construction process , And can handle continuous attributes , It can also process incomplete data . It can be said to be in decision tree classification , Algorithm with milestone significance .

2.  Naive Bayes （NaiveBayes）

Naive Bayesian model is based on the principle of probability theory , Its idea is like this ： Want to classify the given unknown objects , It is necessary to solve the probability of each category under the condition that the unknown object appears , Which is the biggest , It is considered that this unknown object belongs to which category .

3. SVM

SVM In Chinese, it is called support vector machine , English is SupportVector Machine, abbreviation SVM.SVM A hyperplane classification model is established in the training .

4. KNN

KNN Also called K Nearest neighbor algorithm , English is K-Nearest Neighbor. So-called K a near neighbor , That is, each sample can use its closest K A neighbor represents . If a sample , its K The closest neighbors belong to the classification A, So this sample also belongs to classification A.

5. AdaBoost

Adaboost A joint classification model is established in the training .boost In English, it means to promote , therefore Adaboost It is a lifting algorithm for constructing classifiers . It allows us to form a strong classifier from multiple weak classifiers , therefore Adaboost It is also a commonly used classification algorithm .

6. CART

CART Represents classification and regression trees , English is Classificationand Regression Trees. Like English , It builds two trees ： One is a classification tree , The other is the regression tree . and C4.5 equally , It is a decision tree learning method .

7. Apriori

Apriori Is a kind of mining association rules （association rules） The algorithm of , It does this by mining frequent itemsets （frequentitem sets） To reveal the relationship between objects , It is widely used in the fields of business mining and network security . Frequent itemsets are collections of items that often appear together , Association rules imply that there may be a strong relationship between the two objects .

8. K-Means

K-Means Algorithm is a clustering algorithm . You can think of it this way , Finally, I want to divide the object into K class . Suppose that in each category , There was a “ Center point ”, Opinion leader , It is the core of this category . Now I have a new point to classify , In this case, just calculate the new point and K The distance between the center points , Which center point is it near , It becomes a category .

9. EM

EM Algorithm is also called maximum expectation algorithm , It is a method to find the maximum likelihood estimation of parameters . The principle is ： Suppose we want to evaluate parameters A And parameters B, In the initial state, both are unknown , And got it A You can get B Information about , In turn, I know B And you get A. Consider giving... First A Some initial value , So as to get B Valuation of , And then from B Starting from the valuation of , Reevaluate A The value of , This process continues until convergence .

EM Algorithms are often used in clustering and machine learning .

10. PageRank

PageRank It originated from the calculation of the influence of the paper , If a literary theory is introduced more times , It means that the stronger the influence of this paper . Again PageRank By Google It is creatively applied to the calculation of web page weight ： When a page chains out more pages , Description of this page “ reference ” The more , The more frequently this page is linked , The higher the number of times this page is referenced . Based on this principle , We can get the weight of the website .

Algorithm is the soul of machine learning , It is also the most essence part . this 10 The three classical algorithms have won the highest votes in the whole field of machine learning , Some other algorithms behind are basically improved and innovated on this basis . Today, you have a preliminary understanding of the top ten algorithms , You just need to know what you know .

  3、 ... and 、 Three realms of machine learning

1.  Master the algorithm entry and exit

The first level , The algorithm itself is a black box , Be able to master the basic application scenarios of the algorithm without knowing the specific principle of the algorithm （ Supervised 、 Unsupervised ）, And the basic usage scenarios of the algorithm , Be able to switch packets to implement algorithms .

2.  Understand the principles , Flexible tuning

The second level is to understand 、 Master the principle of algorithm , On this basis, we understand the key technologies in the process of algorithm practice 、 Core parameters , It is best to be able to manually implement the algorithm using a programming language , Be able to interpret algorithm execution results , And on the basis of understanding the principle, optimize the algorithm by adjusting parameters .

3.  Achieve mastery through a comprehensive , Design algorithm

The last level , In fact, it is also an algorithm （ Research and development ） The main tasks of the Engineer , That is, it can be combined with business scenarios 、 Own mathematical foundation to carry out targeted algorithm research and development , This part of the work not only needs a solid knowledge of the basic principles of the algorithm , You also need solid programming skills .

summary

Today I made a list of the knowledge you need to master in learning machine learning , Only you know the process of machine learning 、 Algorithm 、 Have a deeper understanding of the principle , You can make better use of , I wish you further and further along the road of machine learning .

The good news ！

Xiaobai learns visual knowledge about the planet

Open to the outside world

 download 1：OpenCV-Contrib Chinese version of extension module 

 stay 「 Xiaobai studies vision 」 Official account back office reply ： Extension module Chinese course , You can download the first copy of the whole network OpenCV Extension module tutorial Chinese version , Cover expansion module installation 、SFM Algorithm 、 Stereo vision 、 Target tracking 、 Biological vision 、 Super resolution processing and other more than 20 chapters .


 download 2：Python Visual combat project 52 speak 
 stay 「 Xiaobai studies vision 」 Official account back office reply ：Python Visual combat project , You can download, including image segmentation 、 Mask detection 、 Lane line detection 、 Vehicle count 、 Add Eyeliner 、 License plate recognition 、 Character recognition 、 Emotional tests 、 Text content extraction 、 Face recognition, etc 31 A visual combat project , Help fast school computer vision .


 download 3：OpenCV Actual project 20 speak 
 stay 「 Xiaobai studies vision 」 Official account back office reply ：OpenCV Actual project 20 speak , You can download the 20 Based on OpenCV Realization 20 A real project , Realization OpenCV Learn advanced .


 Communication group 

 Welcome to join the official account reader group to communicate with your colleagues , There are SLAM、 3 d visual 、 sensor 、 Autopilot 、 Computational photography 、 testing 、 Division 、 distinguish 、 Medical imaging 、GAN、 Wechat groups such as algorithm competition （ It will be subdivided gradually in the future ）, Please scan the following micro signal clustering , remarks ：” nickname + School / company + Research direction “, for example ：” Zhang San  +  Shanghai Jiaotong University  +  Vision SLAM“. Please note... According to the format , Otherwise, it will not pass . After successful addition, they will be invited to relevant wechat groups according to the research direction . Please do not send ads in the group , Or you'll be invited out , Thanks for your understanding ~