#supervised_learning

Types of Machine Learning problems:

Supervised

Un-Supervised

Reinforcement Learning

Supervised Learning: Here, we want to make certain predictions for the future. Hence, we want the machine to learn the previous historical data and forccast for the future for instance, temperature for today. We provide both the input value and output (labels) of historical data such as climate details, humidity etc and along with that output like temperature. Hence, the model can find the derivatives between input and output and generate an equation. eg regression model.

Unsupervised Learning : USL is to simply find out different patterns in data and categorize something Or group something or segment something. Here, we only provide input values not output values. We will only provide the relative features of the class but will not label them. For instance, for identifying a dog, we give long tail, sharp teeth, sharp claws, makes a boow noise etc. But we will not give the label for it. We only expect the machine to make a segregation based on the underlying features. Therefore, Unsupervised Learning does not make any predictions for the future but only makes segregation.

Reinforcement Learning: it is known as reward-based learning. For instance, we have a robot that is learning how to walk. We train the robot that walk straight and if you strike anything on your way like wall or table etc, then turn left or right and move forward or come back. Each time he strikes somewhere, we would point out his mistake and tell him where he is going wrong. Reward him in such a way that if we does not strike anywhere, he provide a rewarded system such that the robot does not commit the same mistake. So this is what Reinforced Learning is all about.

Check out our master program in Data Science and ASP.NET- Complete Beginner to Advanced course and boost your confidence and knowledge.

We are using machines to predict results based on incoming data. Basically, we have three types of ML: a) Supervised Learning b) Unsupervised Learning c) Reinforcement Learning

a) In Supervised Learning (having data with labels), the data we received already has categories. eg csv or excel files, with rows and columns having well-labeled data. In supervised learning scenario, we generally use a classification or regression model. In classification model, we can decide whether a fruit is an apple or mango based on input labels. In regression model, based on inputs, we can predict the outcome. eg., we can select candidates in interview based on inputs labels like qualification, skills, experience etc provided in a resume.

b) In Unsupervised Learning, sometimes we get data that doesn’t have labels or isn’t classified, allowing the model to act without any guidance, which is called unsupervised learning. Here the task of the model is to group together unsorted data according to similarities, patterns, difference etc without any prior training. We can discover inherent groupings in the data, such as grouping customers on the basis of historic purchasing behavior. We can use a clustered Learning model and and Association Rule Learning model. In clustering model, we are grouping customer by their purchasing behavior whereas Association Rule Learning model, we make an association such that we make an inference such that a customer who purchases Product A is most likely to purchase Product B as well and group them together.

c)Reinforced Learning is applicable for skill acquisition or Real time Learning eg learning a game by playing it hundreds of times till it achieves the skill level. it learns the art through trial and error.

Check out our master program in Data Science and ASP.NET- Complete Beginner to Advanced course and boost your confidence and knowledge.

 Supervised vs Unsupervised Learning