[MUSIC] We're not starting from
scratch in this course, and we're going to assume some
background from all of us. In fact, we're going to assume that
you've taken the first two courses in the specialization. This is really fundamental. From the first course, you gained a wide view of what's
possible to do with machine learning. And if you already have that view,
it's okay. But this is something that's going
to be very helpful, as well as facilitating kind of your programming
skills, manipulating data and all that. We're also going to be taking the second
course, where we cover issues like what's a machine learning algorithm, gradient
ascent, over-fitting, validation set and cross-validation, bias-variance trade off,
and regularization. So we'll see if we remember those topics. Go back and review if you don't. If you haven't taken any of the courses
but you know what the techniques are, feel free to jump straight into this course but
we'll assume them throughout the course. Just like the other courses we'll
assume some basic background so you're going to take a few
derivatives here and there. You should know what a vector is
because we're going to use those. And we're going to use
some basic functions, including exponentiation and logarithm. So if you need to take a little
refresher on those basic functions, this is a good time to do it. This is a hands-on course, for every
module you're going to do some programming with real world data and
get real world results. So you should expect that, but that doesn't mean that you're
going to have to do some programming. We've set up an infrastructure makes it
easy for you to do it if you know Python. If you don't know Python, you can catch up
on python pretty quickly just like you did in previous courses or you can implement
in whatever language you want. We don't assume that you use
a particular language, but it would be a lot easier
if you use Python. Unlike this first course in
the specialization we'll use heavy use of GraphLab Create. Because use a good black
box to get started. In this course, we're going to not
really heavily on GraphLab Create. We do suggest that you use SFrames, which is the data manipulation open source
library that was created by Dato, or you can use other libraries like
pandas if you prefer in Python. And there will be some assignments,
where using the pre-implemented algorithm, just try to understand how things behave
before you implement them yourself. For that,
we do suggest you use GraphLab Create, but you can use other libraries
like scikit-learn. It's up to you. It will be easier if you use Python,
and we will give you some starter code and some other things,
but you are welcome to use your own. But the net result of this course is all about you implementing your own
machinery algorithms from scratch. So that's what you should
be prepared to do. Now you will need a computer that
has a little bit of power to it. A laptop should be fine,
64-bit machine will make a big difference. You're going to need access to
the internet to download data sets and everything else, and
to watch these videos. And you have to have the ability to
install Python, maybe GraphLab Create if you choose to use it, and store a few
gigabytes of data on your machine. Just like you did with previous courses. We'll also provide some
other resources for those who don't have their own machines. They can use machines on the web, and
we'll talk about that too in our readings. [MUSIC]