How to distinguish between supervised learning and unsupervised learning
Among the common methods of machine learning, we know that it is generally divided into supervised learning and unsupervised learning. (and of course semi-supervised)
l Supervised learning: Supervised learning, in simple terms, is to give a certain training sample (it must be noted here that this sample is both data and the corresponding result of the data), and use this sample to train to obtain a model (it can be said that A function), and then use this model to map all the inputs to the corresponding outputs, and then make simple judgments on the outputs to achieve the problem of classification (or regression). To make a simple distinction, classification is discrete data, and regression is continuous data.
l Unsupervised learning: Similarly, a sample is given, but this sample only has data, but there is no corresponding result, and it is required to analyze and model the data directly.
For example, when we visit an art exhibition, we know absolutely nothing about art, but after appreciating multiple works, we can also divide them into different factions (such as which are more hazy and which are more realistic, even if we do not know what It is called hazy school, what is called realism school, but at least we can divide them into two categories). A typical example of unsupervised learning is clustering. The purpose of clustering is to group similar things together, and we don't care what the class is. Therefore, a clustering algorithm usually only needs to know how to calculate the similarity to start working [2]
As for the question, I can tell you this. When buying a house, the area of the house and its corresponding price are given for analysis, which is called supervised learning; but given the area and no price, it is called unsupervised learning.
Oversight means that a standard is given as "oversight" (or understood as a limit). That is to say, after modeling, there is a standard to measure your right and wrong; unsupervised means that there is no such standard. After clustering the data, there is no standard to measure it.
The content of the article is personal opinion for your reference. If there are any mistakes, please correct me, thank you.
Special thanks to various great gods on the Internet for selfless sharing and sharing.
refer to:
[1]http://www.zhihu.com/question/23194489
[2]http://blog.csdn.net/warrior_zhang/article/details/41453327
<!--EndFragment-->