Now there are three types of machine-learning:
1. Supervised learni
2. Un-well-trained learning
3. Reinforcement learning
Supervised learning
In this process, a learning-algorithm is given some specific information to get a specific result. If he receives such information in the future, he can provide the desired response. For example, if a teacher shows a lot of examples to his student, then he can take out a general rule or formula from those examples. If a learning algorithm is trained with a sample of a person's hand writing, then the person's computer will be able to recognize it later. If an algorithm is shown with pictures of thousands of different types of cats and dogs then after seeing a picture it can tell that it is not a cat nor a dog. Such machine learning (ANN) by spam filtering in our mailbox. When a user marks an e-mail as spam, he is actually training a learning algorithm, unknowingly.
Spam e-mails are filtered using supervised machine learning.
Examples of some practical applications of supervised learning can be given:
1. Spam filtering
2. Voice Recognition
3. Optical Character Recognition
4. Object detection (computer vision) and more.
Un-well-trained learning
Un-supervised learning is what it says when without knowing the algorithm, completely alienated information, without supervisor or interference. In this process, the algorithm itself is left to be taught by itself. Learning English can be a good example of un-well-trained learning by learning how to hear and hear in England. This method, which can be used to help humans by exploring the unreleased meaning of hundreds of isolated information. The biggest example of this is Google's add system. Google's algorithm with a variety of user information is understood to be about whether this user has interest. And it shows Ed on that.
Special data by selecting unsupported learning algorithm from isolated data.
Some examples are:
1. Netflix movie review system
2. Emamazon Product Reconditionation System etc.
Reinforcement learning
It is half-supervised half-unsolved learning system. That is to say, the algorithm will be given incomplete information in the right way, but the result is shown in the results. At the end of a process, it is correct to tell the correct result, and it is wrong to tell if it is wrong. To make the algorithm for decision-making work that can be changed from time to time, the reinforcement algorithm is implemented for the purpose. An aerial pilot software is allowed to travel repeatedly between aircraft training simulations. If negative feedback is more successful, then positive feedback is given. Elgarydom decides what to do with feedback, thus learning how to operate the plane.
A program in the Rewards-Punishment system can learn what its next step is. Examples of some reinforcement application applications are:
1. Auto driver training
2. Play game
3. Robotics etc.
We've been consistently dependent on this machine learning, unintentionally. Google Assistant on the smart phone, SIRI of i-phone or smart unlock (new technology of iPhone ten) or Windows Cortana is the machine learning technology that is constantly learning about users, updating their policies and becoming a user. Machine prediction is used in weather forecasts, sports predictions. But today, the biggest achievement of machine learning is the loss of world champions in the Japanese game, and in the end of last year, discover new habitable planets. Google's NASA invented a new planet in a jointly organized mission by Google's machine learning software.
There is no alternative to using machine-learning to create artificial intelligence systems for our future people. The use of this technology continues to increase. However, this technology is still not very advanced yet. This technology needs to be improved if people depend on the decision and help given by this technology. For this purpose computer scientists are continuously doing research; Working with Gogal, Amazon, Microsoft, Facebook and other big companies. We hope that in their study we will be more dependent algorithms that will make our life easier and beautiful and will give us the responsibility of our extra work to retire from reading books.