Machine learning model
Hello friends, do you think that what is this machine leaning and what is the model, then do not worry because in today's article, tell about what is machine learning and what is the machine learning model.
 |
| Machine Learning |
If we look at today's world, we see that today everyone in the world is rapidly adopting Machine Learning and Artificial Intelligence. As a result, many industries are waking up to the benefits of machines and computers. By which they decide to repeat the tasks without human intervention. If you want to know about machine learning model in depth then stay with us till the end of this article.
Table Of Content
1) What is machine learning models?
2) What are the different types of Machine Learning?
3) What are the different machine learning models?
4) FAQ
1) What is machine learning models?
 |
| Machine learning Model |
Machine learning is also often referred to as predictive analytics or predictive modeling.
• supervised learning
What happens in Reinforcement Learning is that you do not need a lot of data, what it does is that it learns from its own experience, as a human being is a human being, he keeps on making mistakes. And later learns from those mistakes and moves forward. Reinforcement Learning was designed keeping in mind what actually happens in Reinforcement Learning, for example we have a robot. If we are bringing many pillars in front of the robot, then what will happen that if the robot crosses that pillar then it will get one point, if it is not able to cross the pillar, it will collide and then one point will be deducted. Our machine learning model, which is in the process of this reward, keeps on getting itself trained and its main goal is to get maximum reward. For example, if we see a real life example of this, when we were young, our parents used to tell us that if you get such a number, then we will get you what you want, Reinforcement learning works on this concept.
Coined by American computer scientist Arthur Samuel in 1959, the term 'machine learning' is defined as "the ability of a computer to learn without being explicitly programmed". You must have come to know what is machine learning. Now let us talk about what is machine learning model. A machine learning model is a program that is trained to recognize patterns within new data and make predictions.
2) What are the different types of Machine Learning?
In general, most machine learning model techniques can be classified into supervised learning, unsupervised learning, and reinforcement learning. Of course, not everyone agrees on the exact number or breakdown of machine learning models, but we're offering a summary of the two most common. Every machine learning algorithm organizes into one of three models:• supervised learning
• Unsupervised Education
• reinforcement learning
Supervised learning is the simplest machine learning model to understand, in which an algorithm is given an input dataset, and is rewarded or optimized for accomplishing a set of specific outputs. so that an algorithm can successfully learn from it. This input data is also called training data.We can apply supervised learning models to simple real life problems. For example, we have a dataset with age and height; Then, we can build a supervised learning model to predict a person's height based on their age.
Supervised learning models are classified into two categories:
I) Classification
II) Regression
I) Classification
Classification models are another type of supervised learning technique, in which the classification task is to predict the class or type of an object from a limited number of alternatives. The output variable for classification is always a categorical variable. For example, the classification model can identify whether an email is spam; Will a buyer buy the product or not, etc. There are two types of classification in
○ Binary classification
○ Multi-class classification
○ Binary classification
Here we will discuss in some depth about binary classification. Binary classification is a type of process, in which we divide a given data into two parts. In most binary classification problems, one class has the normal state and the other has the abnormal state.For example, suppose you have been sent two emails, one from an insurance company that keeps sending its advertisements, and the other from your bank regarding your credit card bill. The email service provider will categorize the two emails, and send the first email to the spam folder and the second email to the primary email. This process itself is known as binary classification, because there are two discrete classes, one is spam and the other is primary. So, this is a problem of binary classification.
○ Multi-class classification
Multiclass classification is a type of classification that classifies a model that has more than two classes. A model is labeled for only one class. When we talk about multiclass classification, we have two classes in the dependent or target variable.For example, classification using features extracted from a set of animal images, where the image could be either a dog, a cat or a rabbit. Each image is a model and is labeled as belonging to one of 3 possible classes.II) Regression
Regression criticism is an important concept in the field of machine learning. This critique is a method for modeling the relationship between dependent and independent variables with one or more independent variables. In this critique, once the relationship between the independent and dependent variables is estimated, the results can be predicted.Therefore, regression algorithms always help in forecasting variables such as house prices, market trends, weather patterns, oil and gas prices, etc. The job of a regression algorithm is finding a mapping function so that we can map an input variable of "x" to a continuous output variable of "y". Type of Regression
a) Linear Regression
b) Decision Tree
c) Random Forest
d) Neural Networks
a) Linear Regression - Linear regression is considered one of the simplest and popular machine learning algorithms. It is a method used for predictive criticism. It acts like a regression. Different regression models differ depending on the relationship between the dependent and independent variables they are considering, and the number of independent variables being used. For example, in finance, linear regression can be used to understand the relationship between a business's stock price and its earnings, or to predict the future value of a currency based on its past performance.
b) Decision Tree - A decision tree is a tree that is a non-parametric supervised learning algorithm. Which we can use for both classification and prediction. A decision tree is a flowchart-like structure in which the internal nodes represent the features of the dataset.
For example, in the example below, decision trees learn from data to approximate a sine curve with a set of if-then-else decision rules. The deeper the tree, the more complex the decision rules and the fitter the model.
For example, in the example below, decision trees learn from data to approximate a sine curve with a set of if-then-else decision rules. The deeper the tree, the more complex the decision rules and the fitter the model.
Random Forest - Random Forest is is an algorithm of machine learning, which is based on a supervised learning model. The Random Forest algorithm is used to solve both regression and classification problems. The name Random Forest itself suggests that it is a collection of multiple decision trees. We have already learned what is a decision tree. Nevertheless, let me tell you in short what is a decision tree. A decision tree is a tree-structure classifier.
When a forest is made up of many decision trees, it is called a random forest.
Talking about the real life example of random forest, suppose you want to buy a dream house, you will go to various sources like property dealers, agents etc. Has been You will gather all this information by going to different people, then on the basis of all these you will build or buy your dream house. Will be different After aggregating all these, when you will reach the conclusion that your final result will come on the basis of what the majority of people said. This is your real life example of random forest.
When a forest is made up of many decision trees, it is called a random forest.
Talking about the real life example of random forest, suppose you want to buy a dream house, you will go to various sources like property dealers, agents etc. Has been You will gather all this information by going to different people, then on the basis of all these you will build or buy your dream house. Will be different After aggregating all these, when you will reach the conclusion that your final result will come on the basis of what the majority of people said. This is your real life example of random forest.
● Unsupervised learning -
If we talk about Unsupervised learning, it is understood by the name itself that learning in which no one supervises the model or say that when there is no instructor or teacher to train the model. In the case of supervised learning, the data we had was well labeled data. Whatever work was done there was done in the presence of Supervisor but in case of Unsupervised learning there is no supervisor or teacher to train our machine what is right and what is wrong. Now here the question arises that how will the machine learn from the unlabeled data, then see here what is the task of the machine, how will it short the unclassified information or unlabeled data, how will it short, on the basis of any similarity in it. Or will see the inequality. Through equality and inequality, he will segregate the data or put up separately and there is no supervisor.
There is unlabeled data, no training is provided to it. The machine, which is there, does not get any knowledge before the data, so here it automatically finds the pattern which is in its data, which is the hidden pattern in it, it will find it and according to this works from.
There is unlabeled data, no training is provided to it. The machine, which is there, does not get any knowledge before the data, so here it automatically finds the pattern which is in its data, which is the hidden pattern in it, it will find it and according to this works from.
There are two types of unsupervised learning.
I) Clustering
II) Association Rule
II) Association Rule
I) Clustering-
Within the clustering that happens, we do grouping, whatever set of objects we have, we will group them in a cluster and how will we do this, we will cluster on the basis of what we see similarity in and the objects in which we see less similarity or If there is no similarity, then we make a separate cluster. For example, we have different types of flowers which are row materials. So here we have applied the algorithm of Clustering and that algorithm has made a group or cluster on the basis of the features and characteristics of the flowers, such as rose separately, sunflower separately, etc. and it is also called a subset of the given data.II) Association Rule-
Association rule says that you have a large data set or a set of objects. It will discover you a relationship in those data sets or sets of objects, it will tell you the interesting relationship that what is the relationship between the data set of your variable.
Like for example we take glossary shopping data now here we see that some customer is buying Bread, Butter and Eggs, some customer is buying Eggs, Bread and Milk and some customer is buying Chocolate, Coke and Chips so here What will the association rule do, it will find out what is the relation in the data object, it will find that relation and it has come out on the relation that most of the people take butter with bread, it has come out on this conclusion. This is called the Association rule.
Like for example we take glossary shopping data now here we see that some customer is buying Bread, Butter and Eggs, some customer is buying Eggs, Bread and Milk and some customer is buying Chocolate, Coke and Chips so here What will the association rule do, it will find out what is the relation in the data object, it will find that relation and it has come out on the relation that most of the people take butter with bread, it has come out on this conclusion. This is called the Association rule.
• Reinforcement Learning
Reinforcement learning is the same model of machine learning as I told in the above article, Supervised learning and Unsupervised learning which is a model of machine learning, similarly Reinforcement learning is also a model of machine learning and it is slightly different from Supervised learning and Unsupervised learning Is. As you will remember that in Supervised Learning, we have a lot of data and we have labels available with us, so what do we do that we take the data and labels and put them in the machine learning model and train them. We predict what we do. This is how Supervised learning works, so we do not have labels in Unsupervised learning. By automatically recognizing the pattern, we get the result. As I told you in the above article, now let's talk about how Reinforcement Learning is different from them, then friends, you must have read about AI or ML, then you will have one thing that our AI is very advanced. It has happened so far that when the AI robot was instructed how to play the game of chess, what it did was that it defeated the world champion of chess with that basic knowledge and now let me tell you how it did it. Done with the help of Reinforcement Learning.What happens in Reinforcement Learning is that you do not need a lot of data, what it does is that it learns from its own experience, as a human being is a human being, he keeps on making mistakes. And later learns from those mistakes and moves forward. Reinforcement Learning was designed keeping in mind what actually happens in Reinforcement Learning, for example we have a robot. If we are bringing many pillars in front of the robot, then what will happen that if the robot crosses that pillar then it will get one point, if it is not able to cross the pillar, it will collide and then one point will be deducted. Our machine learning model, which is in the process of this reward, keeps on getting itself trained and its main goal is to get maximum reward. For example, if we see a real life example of this, when we were young, our parents used to tell us that if you get such a number, then we will get you what you want, Reinforcement learning works on this concept.
Q-learning -
Q-learning is an off Policy Reinforcement Learning algorithm, if we say off policy, here according to the situation, the agent will decide which action to perform on which state, the agent does not have any pre-made policy. Because Q-learning is a model free Reinforcement Learning in which the agent only knows what is the set of possible states in the environment and what state it is currently in the environment but it does not know which state But what are the features and rewards he will get on performing the action, what is the transition probability of going from one state to another. As there are 10 states in an environment from S1 to S10, he will know that he can perform the action which is left, right, up and down and he will know that and currently he is on S1 state then it will be known but On performing which action on which state, which state will be reached and what reward will be received, it is not known. Because it is model free Reinforcement Learning algorithm. Here the agent has to learn through experience by interacting with the environment that by performing which action on which state maximum rewards can be obtained.
Q-learning is an off Policy Reinforcement Learning algorithm, if we say off policy, here according to the situation, the agent will decide which action to perform on which state, the agent does not have any pre-made policy. Because Q-learning is a model free Reinforcement Learning in which the agent only knows what is the set of possible states in the environment and what state it is currently in the environment but it does not know which state But what are the features and rewards he will get on performing the action, what is the transition probability of going from one state to another. As there are 10 states in an environment from S1 to S10, he will know that he can perform the action which is left, right, up and down and he will know that and currently he is on S1 state then it will be known but On performing which action on which state, which state will be reached and what reward will be received, it is not known. Because it is model free Reinforcement Learning algorithm. Here the agent has to learn through experience by interacting with the environment that by performing which action on which state maximum rewards can be obtained.
SARSA (STATE-ACTION-REWARD-STATE-ACTION ) -
It is a modified Q learning algorithm where target policy is same as behaviour policy. The two consecutive pairs and the immediate reward received by the agent while transitioning from one state to next state determind the update Q value. So value this method is called SARSA. As target policy is as behaviour policy. SARSA (STATE-ACTION-REWARD-STATE-ACTION) is an on policy learning algorithm.
