unsupervised learning model We look at this image H, which is much similar to supervised learning, but one thing is missing. I think I hope you would have predicted what is that one thing. Here, there is no label. Only the machine learning algorithm takes the training set, which does not have labels, only blank features are given. So, as usual, from the training set, the features are extracted, and the algorithm trains, and the model predicts the new document. It actually clusters the document and adds this new data to the respective cluster or respective group, we can say. So, the prediction here is the likelihood of taking part in a particular group, right? Since there is no label and it may be the goal is to learn a mapping from input to output without any predefined class labels, so we have to find the interesting pattern in the data, how the data are similar to each other. Okay, like once you find out the similarity between the data, you can group the data very easily. Suppose assume that there are a group of students in front of you. There may be a mixed level of heights of students. Suppose you wanted to group the students according to their height, maybe the tallest group and shorter group. So, we have to start with, we have to, what is, see what you will manually, if you do, what you will do? We have to find a person randomly who seems to be taller and another person randomly maybe a shorter and compare the others with both of them, and to whom the new person is much similar, you attach them to that particular group. Right, that is what grouping occurs. That is what interesting. This is what the interesting pattern in the grouping data. So that the unsupervised learning could be written as a set of training for a set of training set, there could be data only, no label, where XI varies from 1 to n. So, there is a D set of training with n number of training examples.