Unsupervised Machine Learning Methods

Name: Unsupervised Machine Learning Methods
Author: Gabriel Popoola

This chapter will introduce unsupervised machine learning concepts and techniques. It will cover Gaussian mixture models. It will also cover various types of…

Preview the first lesson free — get full access to all 3 lessons.

Request a demo Sign in

Course: On-Demand

Welcome to KnowledgeCity's course on machine learning, and supervised machine learning methods. In these lessons, you'll learn about Gaussian mixture models, clustering, including k-means, DB scan, and optics, and manifold learning, including isomap, multi-dimensional scaling, and t-distributed stochastic neighbor embedding. This lesson is going to be on Gaussian mixture models where you'll learn what is a Gaussian mixture model, how are they used and how are they implemented? The Gaussian mixture model is a type of probabilistic model that assumes all points within a data set are generated from a finite number of Gaussian distributions. It can be viewed as a way of clustering your data in such a way that each cluster has a Gaussian distribution. This will make it possible to identify which cluster an unknown point belongs to with a high level of confidence. How are Gaussian mixture models used? Typically, they're used to represent normally distributed subpopulations within an overall population. They also are capable of automatically learning subpopulations without having any indication of which subpopulation a particular point may belong to. Next, we're going to get into how Gaussian mixture models are implemented in Python. This is using the scikit-learn package, and here, I'm going to show a clustering example, and then I'm going to show you how the distributions look for each of the clusters. So all I'm doing here is importing my packages, and then I'm creating a data set of noisy circles, which you'll see exactly what that looks like shortly. So what we have here are two circles. They're just made up of, you know, 5,000 points total. And here, I'm defining the mixture model. And for n components here, it's just the number of components, and so basically what that means is, because I'm using it as a clustering technique here, there are going to be two clusters or two subpopulations that are detected or defined. So now that that's done, I'm going to plot the two clusters. And so, as you can see here, the Gaussian mixture model separated the circles diagonally here, instead of an inner circle and outer circle. And so now that we have this, I just want to show you... A comparison. So, here we have the full population data, and then here we have the subpopulation data. So the idea here is that because this is a Gaussian mixture model, the model did its best so that the data that is blue and the data that's green both have Gaussian distributions. And so to check that, I'm going to plot those distributions. And so, as you can see here, both distributions are the same. In fact, they're just reflections of one another, and they're not perfectly Gaussian, but they do have the general shape of a Gaussian distribution. And so, two things to know about this. One, the data wasn't actually Gaussian distributed, right? And so, what the algorithm did was take the best, you know, split of the data, or, you know, cluster of the data that would get it as close to the Gaussian distribution as possible. If this data was actually Gaussian distributed, or even if we had more points, these plots would look more like the Gaussian distribution curves that we're used to seeing. So this concludes this lesson, and next up, we're going to discuss clustering. Thank you.

Beginner Provider : Gabriel Popoola 3 Lessons · 14m in Arabic, German, English, Spanish, French, Portuguese, Urdu, Chinese

Course Description

This chapter will introduce unsupervised machine learning concepts and techniques. It will cover Gaussian mixture models. It will also cover various types of clustering. This chapter will also introduce the concept of manifold learning and the difference between various types.

What You'll Learn

Understand the core concepts and techniques of unsupervised machine learning
Learn about Gaussian mixture models
Explore various types of clustering, including spectral clustering
Examine manifold learning and the differences between its various types

Key Takeaways

Unsupervised machine learning covers concepts and techniques that include Gaussian mixture models, clustering, and manifold learning.
Clustering can take various types, and spectral clustering is among the methods relevant to this material.
Manifold learning is introduced along with the differences between its various types.
The course addresses machine learning algorithms and data mining methods within an unsupervised learning context.

Frequently Asked Questions

What topics does this course cover?

It introduces unsupervised machine learning concepts and techniques, covering Gaussian mixture models, various types of clustering, and manifold learning, including the difference between its various types.

What lessons are included?

The course includes three lessons: Gaussian Mixture Models, Clustering, and Manifold Learning.

What skills will I gain from this course?

You will build skills in data mining methods, machine learning, machine learning algorithms, machine learning methods, spectral clustering, and unsupervised learning.

What are the learning objectives of this course?

The learning objectives are to learn about Gaussian mixture models, clustering, and manifold learning.