Methods of machine learning.

Methods so as to realize machine learning.
There are various of methods to realize machine learning and in procedure languages we call them algorithms or training methods and nowadays machine learning has three main algorithms the first one is supervised learning and the second one is on supervised learning and the third one is reinforcement learning.Supervised learning is the most frequently used methods of machine learning.And it is talking about the machine training by human beings.And in the area of machine learning we call those data input as training data and every group of training data has a clear tag or result and the target of supervised learning is to build a predictor model and during this progress the learning algorithms divide the predicted result with training data,So as to adjust the Data of the models until it has reached the precise Percentage of the models and the supervised learning in the area of detect the characteristics and progress,Those images and evaluation of credit and the prediction of illegal risk and the prediction of Stokes and so on.So what we talk about the supervised learning the supervised means we have standard answers when answering these practice tasks. So every time we finish the task we get those answers to chat with or not our answer is right if it is wrong so returns to supervise the progress so as to find where is from so as to improve our abilities until the targets are finished. And then we try to get into those tests so as to evaluate our levels.And the training of machine is the same progress, the data used in the training progress are targeted with some tax and answers so as to be trained in the machines and to improve the ability of those machines. And after those training progress, those machines need to take part in those tests so as to face a new data set and to see whether or not those machine acts and performances to give those results and evaluation.The second one is about on supervised learning and it is a statistic method on the earth and those data without tax and tips were discovered through those potential structures in the machine learning system.It intends to find those unknown regulations and rules without outside help so as to divide those data into diverse classifications so as to describe easily to practice those machines, but without answers so as to learn it on their own.And the on supervised learning mainly use in the area of exploration and Discovery of knowledge and they try to find those potential pattern or inner structure of these data input in some situation human beings cannot know the answer of those data set because of some difficulties and cannot take those data so as to have no answers, even though some algorithms supporting those machines so as to train those data.But only if we know about how to calculate the similarity and then we can start on supervised learning algorithms.So we talk about the difference between supervised learning and unsupervised learning we can know the first one is about get the input and output answer so as to calculate and get the algorithms and progress methods of the system so as to get those algorithms and resolution of the problems. And then we get some test data so as to get the precise percentage of the data and some impact features so as to get a clearer system and models of the Problems.And on supervised learning when you talk about whether or not we can get the algorithms without answers and then get those similarity of the same kinds of data. So as to divide those input data into diverse kinds of sections and classifications so as to get diverse kinds of data and divide them into different sides so as to get those unusual and usual classifications and to get the diversity of those input data.And the last one is Reinforcement learning similarity as supervised learning reinforcement learning use computer to get those correct answers From those on learning problems.And different from the supervised learning supervised learning only check those answers right along with the input data but those reinforcement learning reflect those input data to the model itself and the models to adjust itself rightly in real time, so as to evaluate itself and to learn from dose reactions and input data consistently.And reinforcement learning is the modeling of progress. Its main idea is to learn in the environment on it on for those intelligent object.And those intelligent object to analyze all kinds of data input from outside, so as to adjust itself to change its action so as to satisfy and react to the outside environment this is the learning method of exploring and Discovery so as to improve itself by get some mistakes and trying some new methods.So reinforcement learning Defines those machines as Intelligent subject and throw those intelligent subject into the environment and that environment has its own rules but those rules were not set directly and clearly so in the following time give those intelligent object some action rules and then those intelligent object will discover and explore how to win the Competitive actors so as to win.And then after trying so many times, the intelligent object get some methods and rules and regulations so as to be a high level competitor.And then in the environment not every steps were remembered by the intelligent object but it knows how to predict and know the trend of the system and environment and get corresponding actions and reflections.