Our Example

Loading...

En provenance du cours de Princeton University

Networks Illustrated: Principles without Calculus

39 notes

Princeton University

Networks Illustrated: Principles without Calculus

39 notes

What makes WiFi faster at home than at a coffee shop? How does Google order its search results from the trillions of webpages on the Internet? Why does Verizon charge $15 for every GB of data we use? Is it really true that we are connected in six social steps or less? These are just a few of the many intriguing questions we can ask about the social and technical networks that form integral parts of our daily lives. This course is about exploring the answers, using a language that anyone can understand. We will focus on fundamental principles like “sharing is hard”, “crowds are wise”, and “network of networks” that have guided the design and sustainability of today’s networks, and summarize the theories behind everything from the social connections we make on platforms like Facebook to the technology upon which these websites run. Unlike other networking courses, the mathematics included here are no more complicated than adding and multiplying numbers. While mathematical details are necessary to fully specify the algorithms and systems we investigate, they are not required to understand the main ideas. We use illustrations, analogies, and anecdotes about networks as pedagogical tools in lieu of detailed equations. Please note that per Princeton University policy, no certificates, credentials or reports are awarded in connection with this course.

À partir de la leçon

Movie Recommendation on Netflix

One of the perks of having a Netflix subscription is getting recommendations of movies to watch. Behind the scenes, Netflix uses powerful algorithms to determine which will be suggested to each person specifically. In this lesson, we will take a look at the main ideas behind these algorithms.

Netflix Timeline5:21

Video Streaming5:19

Recommendation is Everywhere3:24

Netflix Recommendation System5:20

Netflix Prize: Logistics7:38

Netflix Prize: The Competition4:33

Our Example3:48

Raw Average7:13

User-movie Interactions7:00

Baseline Predictor6:57

Cosine Similarity9:18

Similarity Values10:58

Leveraging Similarity8:30

Neighborhood Predictor4:44

Performance of Different Methods3:30

Copy of Cosine Similarity9:18

Rencontrer les enseignants

Christopher Brinton
Lecturer
Electrical Engineering
Mung Chiang
Professor
Electrical Engineering

So you're probably wondering what was in Bell Core's pragmatic chaos or the

Ensembles algorithm that made them work so well.

Well, turns out that there's really hundreds and hundreds ingredient

algorithms that are really blended together in order to make it work as well

as it did. So here we're not going to go through all

that. We're going to illustrate the main ideas.

Again, there's hundreds of ingredient algorithms.

We're going to focus on certainly two of the main ideas that are, kind of form the

baseline in order to get there. Two of the main themes that they used,

and, the first one of them is the baseline predictor.

It's, no pun intended. I said before, forming the baseline.

This does form the baseline for what's called the neighborhood model, and so

basically, in the baseline predictor, we're going to just look at trying to use

some sort of biases, HDL users, in order to make a prediction.

So this is going to be something like user movie biases.

So, is this user an easy or harsh user, is the movie well received, not well

received and modify the predictions accordingly.

And then the neighborhood method, we're actually going to do sort of,

similiarities. So we're going to say like, user To user

or movie to movie similarities. So for instance if Charlie and Bob tend

to like all the same movies then we can say that if Charlie's going to like some

movie that Bob probably likes it too. then at the same time movies could be

like one another so if you may have two different movies two different Disney

movies for instance Like Aladdin and the Lion King, right, and users tend to

either like or not like them, so there, you can therefore say that if one user

likes one movie he's probably going to like the other one.

And, so, now let's look at our example that we're going to be using to

illustrate this. We'll stick to one example just to make

it easier. And so, in our example, we have six users

which we label A through F down the rows again, users down the rows and, we have

five movies one through five, which we use Roman numerals to denote.

the reason I did letters and then Roman numerals was so you didn't get confused

with them and the actual rating values which will use just, numbers, one through

five to do that. And so this, this table is a lot more

dense than is necessary. We said before that Netflix really was a

very sparse data set, alright so many entries are just not filled in.

out of the even a hundred million entries that there are, there's just many, many

more users than movies than there are entries.

When you consider all the different pairs that could be filled out on the table.

And, so, in this, in this case right here, you know?

We have all, except for 5 of these, filled.

So, you know? Really, we have, you know, quite a lot of

the data. and that's not going to be the case in

real life. But it'll, this will suffice to

illustrate the, the main ideas. And,

So we're going to section our data as well.

again, we have our training data, which are the.

the what the numbers right here that aren't that are just written in black so

5, 4, 2, 4, 4 and so on. Then the test data we've, we've unbolden

them and colored them blue. So for instance 4 is in the test set.

2 is in the test set. 5, this 3 value and this 4 value over

here they're all in the test set. So what users see rates movie 1 we do

know the value. But we're going to hold that off and make

that part of the test set in similar with what user D, rates would be too.

So, we're not going to use those when rating algorithm, we're just going to use

them when we compare in RMSC in the predictions.

And then there's unknown ratings, which we just don't have, because the users

haven't watched them or rated the movies yet.

For instance, user E and movie 2, he hasn't rated that.

user c, movie 3, this hasn't been rated. And so on.

Explorer notre catalogue

Rejoignez-nous gratuitement et obtenez des recommendations, des mises à jour et des offres personnalisées.

Coursera propose un accès universel à la meilleure formation au monde, en partenariat avec des universités et des organisations du plus haut niveau, pour proposer des cours en ligne.

© 2018 Coursera Inc. Tous droits réservés.

Télécharger dans l'App Store

Disponible sur Google Play