Machine learning what is supervised learning and unsupervised learning

machine learning is divided into two types: supervised learning and unsupervised learning . Next I'll give you a detailed introduction to the concepts and differences between the two methods.

Supervised Learning (supervised learning): train through an existing training sample (known data and its corresponding output) to obtain an optimal model, and then use this model to map all new data samples to the corresponding output, The optimal model also has the ability to classify the unknown data by simply judging the output result and realizing the purpose of classification. in society, we were taught by adults at a very young age this is a bird ah, that is a pig ah, this is watermelon, pumpkin, this can eat, that can't eat ah and so on, our eyes see these scenery food is machine learning input, the big people tell us the result is output, over time, when we see more , the adults say more, we will form an abstract model in the brain, the next time in the absence of adult reminders to see the villa or the building, we can identify it is a house, can not eat, the house itself can not fly and other information. When you go to school, the teacher teaches literacy, mathematical formulas, English words, and so on, and we can separate and identify them when we meet next time. This is supervised learning, which is ubiquitous in our lives.

Unsupervised Learning (unsupervised learning): We do not have any training data samples, we need to model the data directly. For example, we go to a painting exhibition, we know nothing about art, but after admiring a lot of works, after we face a new work, at least we can understand the work is what faction, such as more abstract or more realistic, although not very clearly understand the meaning of this painting, But at least we can divide it into which category. For example, we are in the movie theater, for we have not learned the relevant film art knowledge, we may not know what is a good movie, what is a bad movie, but after watching a lot of movies, we have a latent understanding of the film in the Brain, When we sit in the cinema and watch the new movie carefully, the brain will have an evaluation of the film: Why the film is so bad, the whole story line is chaotic, not clear, than I have seen before the film is far away, character's character has not shown, the key is the film theme is also biased; The film is really good, the plot and character are very vivid, and the scene is very lifelike, the protagonist's strength performance coupled with his innate melancholy look at the characters to live.

Give us an example of unsupervised learning. Ancient times, our ancestors hunted to eat meat, they were not experienced before, when someone with a very coarse stone to cut the animal's skin, found it difficult to separate the skin, but someone with a very thin stone to cut, found more easily than others to separate animal fur, so, the next day, the third day 、......, They knew they needed to look for thinner pieces of stone to cut. These are unsupervised learning ideas, no outside experience and training data samples provided to them, completely on their own.

Summarize

Thinking back to the two methods of supervised learning and unsupervised learning, perhaps a lot of people would think that anything someone teaches is certainly good, ah, all supervised learning is more convenient and efficient, most of the situation is true, but if some circumstances such as the inability to provide training data samples or provide training data samples of the cost is too high, Perhaps we should adopt a strategy of unsupervised learning. The typical examples of supervised learning are decision trees, neural networks, and disease monitoring, while unsupervised learning is a long-backgammon and clustering.

Machine learning what is supervised learning and unsupervised learning