[machine learning] what is machine learning?

What is machine learning ？

Provides two definitions of machine learning . Arthur · Samuel （Arthur Samuel） Describe it as ：“ Research areas that allow computers to learn without explicit programming .” This is an older informal definition .

Tom Mitchell Provides a more modern definition ：“ If the computer program is T Performance on tasks in （ With P To measure ） With experience E Improve , From experience E Learn about a certain type of task T And performance metrics P. ”

for instance ： Playing chess

• E = The experience of playing checkers many times ​​ Examination

• T = The task of playing chess .

• P = The probability that the program will win the next game .

Generally speaking , Any machine learning problem can be classified into one of two categories ： Supervised learning and unsupervised learning .

Supervised learning

In supervised learning , We got a data set , And we already know what our correct output should look like , And know that there is a relationship between input and output .

Supervised learning problems are divided into “ Return to ” and “ classification ” problem .

• In the regression problem , We try to predict the results in continuous output , This means that we try to map input variables to some continuous function .
• In the classification problem , We try to predict results in discrete outputs . let me put it another way , We try to map input variables into discrete categories .

Example 1：

Given data on the size of houses in the real estate market , Try to predict their prices . Price as a function of scale is a continuous output , So this is a question of regression .

We can turn this example into a classification problem , Rather than let our output about whether the house “ The selling price is higher or lower than the asking price ”. ad locum , We divide houses into two separate categories according to price .

Example 2：

(a) Return to —— Give a picture of someone , We have to predict their age based on a given picture .

(b) classification —— Given a patient with a tumor , We must predict whether the tumor is malignant or benign .

Unsupervised learning

Unsupervised learning enables us to solve problems with little knowledge of what the outcome should be . We can deduce the structure from the data , And we don't necessarily know the influence of variables .

We can deduce this structure by clustering the data based on the relationship between variables in the data .

For unsupervised learning , There is no feedback based on the predicted results .

Example ：

• clustering ： collect 1,000,000 Two different genes , And find a way , These genes are automatically grouped to some extent similar or through different variables （ For example, life 、 Location 、 Roles, etc ） In related groups .

• Nonclustering ：“ Cocktail party algorithm ”, Allows you to find structures in a chaotic environment .（ That is to identify personal voice and music from the voice network at the cocktail party ）