How do I get started with machine learning? What are the methods of machine learning?

The best way to get started with machine learning is to learn theory and learn together. While looking at the corresponding theoretical derivation, we also read and practice the classics**. So, to get started faster, I recommend that you know a little bit of MATLAB or Python.

To be honest, MATLAB and Python are definitely not recommended for high-end machine learning, but if you want to get started with machine learning quickly, then these two languages are definitely excellent choices.

Step 1. With that in mind, you're ready to start looking at machine learning. I see a lot of people recommend Elements of Machine Learning.

I would say, is it really appropriate for you to want a person with zero foundation to read this book???

So, what I recommend is Machine Learning in Action, which is in Python, which is in English. Of course, if you feel that English is a completely insurmountable hurdle for you, (although I recommend that anyone who does technology must at least understand English) there is now a Chinese version, called "Machine Learning Practice".

This book goes through the basic algorithms of machine learning with as few formulas as possible, and it is very clear, and more importantly, he combines formulas with **. So, your machine learning isn't so abstract anymore, you know how the formulas in the algorithm translate into **.

So, as a first step, you can patiently read the book to the end. Anyway, at that time, I knocked the ** in the book myself, although **yes**, you can also choose to just understand it. But I still recommend knocking it yourself and running it, so that you will get a different experience.

Step 2. Take a machine learning course from Andrew Ng on Coursera. This course has benefited many beginners in machine learning, not only because of the comprehensive course and the content from the basic to the deep.

In short, be patient to go through this course once or even a few times.

Step 3. At this point, you already have a good understanding of many simple algorithms for machine learning, but you may not have a big picture. Therefore, I suggest you take a look at these two Chinese textbooks.

Mr. Zhou Zhihua's watermelon book "Machine Learning" and Mr. Li Hang's "Statistical Learning Methods".

In the cutting-edge section and the first to third steps, if you can follow these steps down, getting started is definitely okay.

Machine LearningMachine learning is a multi-disciplinary discipline that has emerged in the past 20 years, involving probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory and other disciplines. Machine learning theory is the design and analysis of algorithms that allow computers to "learn" automatically. Machine learning algorithm is a kind of algorithm that automatically analyzes and obtains rules from data and uses the rules to perform ** on unknown data.

Because there are a large number of statistical theories involved in learning algorithms, machine learning is particularly closely related to statistical inference, also known as statistical learning theory. In terms of algorithm design, machine learning theory focuses on achievable and effective learning algorithms.

Commonly used methods in machine learning are:

1) Inductive learning.

Symbolic inductive learning: Typical symbolic inductive learning includes example learning and decision tree learning.

Functional inductive learning (discovery learning): Typical functional inductive learning includes neural network learning, example learning, discovery learning, and statistical learning.

2) Deductive learning.

3) Analogical learning: Typical analogical learning has case (exemplary) learning.

4) Analytical learning: Typical analytical learning includes interpretation learning and macro operation learning.

Expand the information of the exhibition:

Common algorithms for machine learning:

1. Decision tree algorithm.

Decision trees and their variants are a class of algorithms that divide the input space into different regions, each of which has its own parameters. The decision tree algorithm makes full use of the tree model, and the root node to a leaf node is a classification path rule, and each leaf node symbolizes a judgment class. First, divide the samples into different subsets, and then divide and recursively until each subset gets the same type of samples, starting from the root node, to the subtree and then to the leaf node, you can get the ** category.

This method is characterized by its simple structure and high efficiency in processing data.

2. Naive Bayesian algorithm.

The Naive Bayes algorithm is a classification algorithm. It is not a single algorithm, but a series of algorithms that all share the common principle that each feature being classified is independent of the value of any other feature. Naive Bayesian classifiers consider each of these "features" to contribute probabilities independently, regardless of any correlation between features.

However, features are not always independent, which is often seen as a disadvantage of naïve Bayesian algorithms. In short, the Naive Bayes algorithm allows us to give a set of features to a class using probability. Compared to other common classification methods, naïve Bayesian algorithms require little training.

The only work that must be done before proceeding is to find the parameters of the individual probability distribution of the feature, which can usually be done quickly and deterministically. This means that even for high-dimensional data points or a large number of data points, naïve Bayesian classifiers can perform well.

3. Support vector machine algorithm.

The basic idea can be summarized as follows: first, the space should be high-dimensional by a transformation, which is of course nonlinear, and then, the optimal linear classification surface should be taken in the new complex space. The classification function obtained in this way is formally similar to a neural network algorithm.

Support vector machine (SVM) is a representative algorithm in the field of statistical learning, but it is very different from the traditional way of thinking, which is to input space and improve dimensions to shorten the problem, so that the problem is reduced to a linear and separable classical solution problem. Support vector machine is applied to spam recognition, face recognition and other classification macro shouting problems.

There are several main methods of machine learning:

Supervised learning: Supervised learning is one of the most common methods in machine learning, in which the system is given a set of sample data with known inputs and outputs, and the system needs to learn a function that can produce the correct output based on the given input.

Unsupervised learning: Unsupervised learning is another common method in machine learning. In unsupervised learning, the system has only input data and no output data. The system needs to learn a function that automatically classifies the input data.

Semi-supervised learning: Semi-supervised learning is a hybrid approach to supervised and unsupervised learning. In semi-supervised learning, the system is given a part of the sample data of known inputs and outputs and a part of the unknown input data, and the system needs to use the known sample data to learn a function, so that the function can produce the correct output based on the unknown input data.

Reinforcement learning: Reinforcement learning is a learning method based on environment and feedback, in which the system learns the optimal strategy through continuous interaction.

Clustering: Clustering is an unsupervised learning method in machine learning, which aims to divide data into different groups, so that the data similarity within the group is maximized, and the data similarity between groups is minimized.

Dimensionality reduction: Dimensionality reduction is an unsupervised learning method in machine learning that aims to reduce the dimensionality of the data so that it is easier to analyze.

Deep learning: Deep learning is a neural network-based learning method in machine learning that mimics the human brain by building multiple layers of neural networks. Deep learning has a wide range of applications in computer vision, speech recognition, natural language processing, and other fields.

Recurrent neural network: Recurrent neural network is a special deep learning method, which processes sequential dead bird data recursively, and has a wide range of applications in natural language processing, speech recognition and other fields.

Bayesian Learning: Bayesian learning is a learning method based on probability theory and statistics, which uses Bayesian theorem to learn and **.

Statistical Learning Methods: Statistical learning methods are a class of machine learning methods based on statistical theory, which uses statistical models and optimization algorithms to learn and learn. Including linear regression, logistic regression, naive Bayes, etc.

These methods have their own characteristics and scope of application, and in practical application, the appropriate method should be selected according to the specific situation of the problem.