你是否好奇数据可以告诉你什么？你是否想在关于机器学习促进商业的核心方式上有深层次的理解？你是否想能同专家们讨论关于回归，分类，深度学习以及推荐系统的一切？在这门课上，你将会通过一系列实际案例学习来获取实践经历。在这门课结束的时候，

Loading...

來自 University of Washington 的課程

机器学习基础：案例研究

8143 個評分

At Coursera, you will find the best lectures in the world. Here are some of our personalized recommendations for you

你是否好奇数据可以告诉你什么？你是否想在关于机器学习促进商业的核心方式上有深层次的理解？你是否想能同专家们讨论关于回归，分类，深度学习以及推荐系统的一切？在这门课上，你将会通过一系列实际案例学习来获取实践经历。在这门课结束的时候，

從本節課中

Recommending Products

Ever wonder how Amazon forms its personalized product recommendations? How Netflix suggests movies to watch? How Pandora selects the next song to stream? How Facebook or LinkedIn finds people you might connect with? Underlying all of these technologies for personalized content is something called collaborative filtering. <p>You will learn how to build such a recommender system using a variety of techniques, and explore their tradeoffs.</p> One method we examine is matrix factorization, which learns features of users and products to form recommendations. In an iPython notebook, you will use these techniques to build a real song recommender system.

- Carlos GuestrinAmazon Professor of Machine Learning

Computer Science and Engineering - Emily FoxAmazon Professor of Machine Learning

Statistics

[MUSIC]

Okay, so let's take these ratings that we were just talking about.

And instead of talking about them for a specific combination of a movie and

a user, let's talk about how we can think about representing our predictions over

the entire set of users and movies.

And to do this, we're gonna need a little bit of linear algebra.

So in particular, if we're looking at the score that we're giving

to a specific movie, v, for a user, u.

So this is rating [BLANK AUDIO]

predicted for user u and movie v.

Well, what did we say that was?

We said we took the user vector Lu, and we took the movie vector Rv,

and we did this element-wise product, and

summed over the elements of that product.

So I'm gonna denote that just with this little notation here.

So these little braces here mean we're gonna take the vector Lu and

the vector Rv, do an element-wise product, and sum.

Okay, so what that represents is we're taking a row

of a big matrix L, so there's a row that has a vector Lu.

So let me write this a little bit more largely here,

so this row Lu is the vector that we talked about before.

With how much that user likes different things like action,

romance, and so on, and

then we take a movie vector, Rv.

So again, we'll write it more largely here.

So Rv is indexed over the same set of genres, or

topics, for the movie, and has some set of entries.

And in this matrix notation,

if you're familiar with matrix multiply, to get the entry.

So this is the uth row here, and the vth column here.

If we multiply these together, we get the uvth entry of this resulting matrix.

So if you're not familiar with this kind of thing, that's okay.

We're gonna talk about it in a lot more detail.

In subsequent courses, but for those who are,

this representation is a very compact way to take all of the vectors for

all of the users, so we're stacking up all of the user vectors,

and stacking up all of the movie vectors.

And through this representation,

we end up with an entire matrix which is just like we were writing before.

It's a users by movies matrix.

So all of the users that appeared here [BLANK

AUDIO] are appearing here, and all of the movies

that appeared as columns here are appearing as columns in this matrix.

And each individual entry, again, is a combination of a specific

row of this L matrix and a specific column of this matrix R here.

[MUSIC]