Supervised vs. Unsupervised Machine Learning

Loading...

Del curso dictado por IBM

Advanced Machine Learning and Signal Processing

212 calificaciones

Explorar los videos relacionados

En Coursera, encontrarás las mejores conferencias del mundo. Aquí te mostramos algunas de nuestras recomendaciones personalizadas para ti

IBM

Advanced Machine Learning and Signal Processing

212 calificaciones

Curso 2 de 4 en SpecializationAdvanced Data Science with IBM

>>> By enrolling in this course you agree to the End User License Agreement as set out in the FAQ. Once enrolled you can access the license in the Resources area <<< This course, Advanced Machine Learning and Signal Processing, is part of the IBM Advanced Data Science Specialization which IBM is currently creating and gives you easy access to the invaluable insights into Supervised and Unsupervised Machine Learning Models used by experts in many field relevant disciplines. We’ll learn about the fundamentals of Linear Algebra to understand how machine learning modes work. Then we introduce the most popular Machine Learning Frameworks for python Scikit-Learn and SparkML. SparkML is making up the greatest portion of this course since scalability is key to address performance bottlenecks. We learn how to tune the models in parallel by evaluating hundreds of different parameter-combinations in parallel. We’ll continuously use a real-life example from IoT (Internet of Things), for exemplifying the different algorithms. For passing the course you are even required to create your own vibration sensor data using the accelerometer sensors in your smartphone. So you are actually working on a self-created, real dataset throughout the course. If you choose to take this course and earn the Coursera course certificate, you will also earn an IBM digital badge. To find out more about IBM digital badges follow the link ibm.biz/badging.

De la lección

Setting the stage

Linear algebra5:53

High Dimensional Vector Spaces2:13

Supervised vs. Unsupervised Machine Learning4:18

Conoce a los instructores

Romeo Kienzler
Chief Data Scientist, Course Lead
IBM Watson IoT
Nikolay Manchev
Data Scientist
IBM EMEA Data Science

So, imagine you have a corpus of data,

and if you can assign a label,

or a flag, or a value to each record in your data,

then you can supervise machine learning.

So, let's consider this table here.

That's table from hypothetical production machine,

where we measure max temperature,

min temperature and max vibration.

Note that those are already aggregations and asperity.

So, asperity is our label.

So, given temperature and vibration data,

we want to predict asperity.

This is a labeled dataset,

and that we can use for supervised machine learning.

Because using our dataset plus the label,

we can train an algorithm in order to predict the future for unlabeled data.

So, let's consider unlabeled new data.

You fit that into our algorithm,

and the algorithm will give us back the label or the value.

So imagine we're giving the algorithm the value of 35, 35 and 12,

then the algorithm will predict a value for asperity of 0.32.

This is called regression.

So, now imagine our label is not a continuous value,

it's a binary value or at least a discrete value.

So, in this case,

the algorithm won't predict a number,

it will predict a class variable zero or one,

a broken yes or no.

This is called classification.

If you have only two classes,

it is called binary classification.

And if you have more than two classes,

you have multi-class classification. That's basically all.

So supervised learning, is either regression,

value predicted continuous value,

or classification, and there you have two different ways of doing it.

You have binary classification,

if you have only two labels or two target values,

or its multi-class classification if you have multiple.

But now, let's imagine you don't have a label to your data,

and that's unfortunately true in most of the cases.

So, I guess around 80-95% of all those data is unlabeled, maybe even more.

So, what can we do there?

So, let's have a look at our three dimensional dataset.

If you pluck those three dimensions,

we can obviously see without a label that there are

three clusters That means that the machine has learned, we have learned something.

So, we learned from the data by just plotting it that

the data is centered around three clusters.

Obviously, for high dimensional data,

we cannot simply plot it but there are

[inaudible] which are taking care of this. So, no worries here.

This process is called clustering.

In contrast to drawing lines,

which is called classification.

Imagine, we have more than three dimensions,

five in this case,

we cannot simply plot.

So, there are two ways of doing that.

We can use a clustering algorithm,

which we will cover later or we use a process called dimension reduction,

which we will also cover later.

So, dimension reduction basically compresses our dataset to

a smaller number of dimensions without losing too much information.

That's a very cool topic and we will also cover it later.

So in summary, just to give you an overview.

Machine Learning contains supervised machine learning and unsupervised machine learning.

Supervised machine learning contains regression,

where you predict the continuous value and classification,

where you predict a discrete value.

In classification, you either predict two values,

this is called binary classification or more than two values,

this is called multi-class classification.

On the other hand,

in unsupervised machine learning,

you have clustering and dimension reduction.

But we will cover all those later in the course.

Explora nuestro catálogo

Inscríbete de manera gratuita y obtén recomendaciones personalizadas, actualizaciones y ofertas.

Coursera brinda acceso universal a la mejor educación del mundo, al asociarse con las mejores universidades y organizaciones, para ofrecer cursos en línea.

© 2019 Coursera Inc. Todos los derechos reservados.

Descargar en la App Store

Consíguelo en Google Play