@J J I didn't, but I actually watch quite good amount of educational and informational videos, out of curiosity, of course, so I can totally see why it landed in my recommended section.
Just rewatching this for the 10th time because my applications of linear algebra class is about to go over this algorithm tomorrow. I’m pretty excited.
not sure whether this does work. There will always be a character/person who endorse noone. But in the limit it will be ranked 100% then and the rest 0 I guess.
@@ericb.4385 if a character doesn't "endorse" any other character, then that would imply that they don't have a weakness. If that is true, then everybody should pick that character to win.
Alessandro # I’ll do you one better : I just fucking read this EXACT, IDENTICAL example of linear algebra applied to real life IN A PAPER BOOK *yesterday* , and then I have this video recommended, from this channel that I’ve subscribed to a while ago; what the fuck ?
How weird is it that I was just about to start the chapter of Graphs in my Data Structures course and before that this video out of nowhere gets recommended to me
This was an amazing video. I was wondering if you can make a video on nuclear engineering. That could explain to anyone who is interested in the field what to expect. Similar to you vid on electrical engineering or civil engineering. Also thank you if your actually reading this :)
can you do a video about data science.I mean not as a major,just a few details about the role data scientist and what classes one should take and what to expect
Why is there music in this video? At first I didn't pay attention to it, but around 8 minutes in, I started to wonder where's this sound coming from, took of my headphones to locate the source of distraction. But it was in the video! Couldn't keep my focus on the explanation after that, all I could hear was that disturbing music 🙄
is this one of markovs? from the looks of it, its about frequency value, as you multiply your adding one loop or one frequency, as they are all in percent form, it will eventually converge. this is good and all, but its beefy in computation, in coding terms, more computation = slower performance. I like the markov arm bandit more as they use less computation, but its more suitable for A/B testing but hey as they say there are a lot of ways to solve a problem.
Great explanation & video, I'm glad YT recommended it. I'm guessing the algo did that b/c I've been watching Free Code Camp's 8hr "Graph Theory" video.
But let’s talk about what the user really wants/likes, why he clicks on a link and the patterns in their clicking. (1) The value of a page is not a function of the page per se. It is a function of the page and what the person wants to do with it. A search for GPS may want to learn how they work, who sells them or what models are available. There is no single rank of a page. (2) A user clicks on a link based only on what he sees - the URL and its surrounding text. The whole point of PageRank is to not provide users with links that they find are worthless. Only after they see the page do they know that - but their click has already been counted. It is only registering the appeal of the URL and what the text surrounding it says. (3) The process of multiplying the click probability matrix repeatedly (raising it to a power) forgets that people generally click a given link only once. Whoever paid $1B for this algorithm should ask for his money back.
why are using markov chains as with the probability of transition as column vectors instead of row vectors?? first time see them like this and is very confusing.
In your video about the uses of matrix math, you mentioned that the eigenvectors of a markov matrix represent a "steady state" of the system. Does that mean that you could think of the matrix representing the links as a markov matrix?
I have a question if its calculating the average time spent on each sight after an infinite amount of clicking wouldnt it rate "traps" way to high? Like if I have a website A and B that only link each other and nothing else. And then I have say 10 other nodes that have a good healthy web between them. If even 1 of them links A or B then after an infinite amount of clicking one would always eventually get stuck in the ABABABAB loop so these would both get a final rank of 0.5 and all other a rank of 0 right?
I wonder how much of this still holds up now that the algorithm is a million times more complicated Also, how does this work when integrating it with a search algorithm?
If a node is isolated then all entries in one of the columns of the matrix become zero (it's no longer a markov matrix). If you raise that matrix to a large power then its entries go to zero which means the ranks do as well.
Using the exact algorithm I mentioned it would be. Of course that’s not how it really works and I was just saying how the video doesn’t mention how that’s accounted for.
@@zachstar I mean if A is somehow a nilpotent matrix the ranks would go to zero. However in a general case with an isolated node that would, like shown, result in a row and column with only zeros. This doesnt have to be a nilpotent matrix. Or plays the fact that there is no self linking allowed a role such that the matrix ist nilpotent ?
@@zachstar But that's what I'm saying that using the exact algorithm not all ranks will be 0, just the rank of the disconnected node, and the ranks of the other ones will add up to less than 1 but not 0.
video about algorithms
gets recommended to everyone
Lol
Skynet would like to introduce itself...
algis
@J J I didn't, but I actually watch quite good amount of educational and informational videos, out of curiosity, of course, so I can totally see why it landed in my recommended section.
Very crispy introduction to Google's Page rank algorithm. I think this is one of the best explanation about Page rank algorithm. You ... Rockzzz....
Don't forget to support him on Patreon.
@@MykolaDolgalov yesssss
Wasn't aware of such a good application of Markov chain. Great video.
Purbesh Mitra beat me to it
Looks like *Somebody* just took linear algebra
Me too bud
A endorsed B, but B endorsed C and D. What a jerk
I was legitimately sad when the video ended
Great work as always❤❤
This channel is very outstanding and gives an in depth explain into a mathematics behind the algorithm
Make more such videos on applied mathematics
Just rewatching this for the 10th time because my applications of linear algebra class is about to go over this algorithm tomorrow. I’m pretty excited.
I wanted to use this method for ranking characters in a video game where an "endorsement" was an indication that the character was weak to another.
Good idea
Wait this would be the number 1 way to do tier lists. Damn man thanks for the idea. Imma start doing that right now.
@@Brettlaken thanks.
not sure whether this does work. There will always be a character/person who endorse noone. But in the limit it will be ranked 100% then and the rest 0 I guess.
@@ericb.4385 if a character doesn't "endorse" any other character, then that would imply that they don't have a weakness. If that is true, then everybody should pick that character to win.
I remember doing this for my Probability project on Markov Chains. I now see where it comes in handy.
The youtube algorithm recommended a video about algorithms. The singularity is near.
My linear algebra teacher literally starts today's lessons speaking about it... wtf 😂
Alessandro #
You have such nice teachers
Alessandro #
I’ll do you one better :
I just fucking read this EXACT, IDENTICAL example of linear algebra applied to real life IN A PAPER BOOK *yesterday* , and then I have this video recommended, from this channel that I’ve subscribed to a while ago; what the fuck ?
@@nicholasleclerc1583 probably majorprep during his degree in engineering developed some sort of psychic power lol
@@alessandromestri9004
Hehe, more like hy majored in supernatural arts or smth like that
Your explanation blew my mind. Good job.
One of the best explanations of the Google Page Rank algorithm. Must watch video for SE webmasters.
How weird is it that I was just about to start the chapter of Graphs in my Data Structures course and before that this video out of nowhere gets recommended to me
I love this channel for a reason
This was an amazing video. I was wondering if you can make a video on nuclear engineering. That could explain to anyone who is interested in the field what to expect. Similar to you vid on electrical engineering or civil engineering. Also thank you if your actually reading this :)
Neatly explained. Great job.
I never knew I needed this
this channel is hecka cool
I love your channel! It's the best!
I'm guessing 'long time' implies, as t -> infinity. is dope, very dope indeed!
Thanks for these top notch videos that you do
Very Informative!
can you do a video about data science.I mean not as a major,just a few details about the role data scientist and what classes one should take and what to expect
Thanks man. You nailed it!!
Best video on page rank algorithm.It would be better if you change the thumbnail.It would attract more viewers.
This is matrix math right
Why is there music in this video? At first I didn't pay attention to it, but around 8 minutes in, I started to wonder where's this sound coming from, took of my headphones to locate the source of distraction. But it was in the video! Couldn't keep my focus on the explanation after that, all I could hear was that disturbing music 🙄
is this one of markovs?
from the looks of it, its about frequency value, as you multiply your adding one loop or one frequency, as they are all in percent form, it will eventually converge.
this is good and all, but its beefy in computation, in coding terms, more computation = slower performance.
I like the markov arm bandit more as they use less computation, but its more suitable for A/B testing but hey as they say there are a lot of ways to solve a problem.
Thanks! Super helpful video
Wow, this video was beautifully detailed. Great video as always.
Great explanation & video, I'm glad YT recommended it. I'm guessing the algo did that b/c I've been watching Free Code Camp's 8hr "Graph Theory" video.
Wait what??? Ima check that out ASAP
Best fricking explanation ever!!!
8:27 will it be a concern if the pagerank is still the same after a long run?
OMG! So concise!
I love the Dexter Polytopes. The grid has a lovely place. The way the lord of the rings worked this out always amazes me.
that's awsome! thanks very much
great video explain the topic clear
Sorry, but I couldn't understand it. Are A B C D are separate sites? Or one site linking pages one to another? :(
what is an outgoing link? is it like a source ? what happens if there are no outgoing links?
But let’s talk about what the user really wants/likes, why he clicks on a link and the patterns in their clicking. (1) The value of a page is not a function of the page per se. It is a function of the page and what the person wants to do with it. A search for GPS may want to learn how they work, who sells them or what models are available. There is no single rank of a page. (2) A user clicks on a link based only on what he sees - the URL and its surrounding text. The whole point of PageRank is to not provide users with links that they find are worthless. Only after they see the page do they know that - but their click has already been counted. It is only registering the appeal of the URL and what the text surrounding it says. (3) The process of multiplying the click probability matrix repeatedly (raising it to a power) forgets that people generally click a given link only once.
Whoever paid $1B for this algorithm should ask for his money back.
Very helpful
great explanation thanks
why are using markov chains as with the probability of transition as column vectors instead of row vectors?? first time see them like this and is very confusing.
I think it is to do with linear solver (some version of gmres) that will be used to rank.
So elegant
In your video about the uses of matrix math, you mentioned that the eigenvectors of a markov matrix represent a "steady state" of the system. Does that mean that you could think of the matrix representing the links as a markov matrix?
Exactly. And the final ranks would be the eigenvector of that matrix.
Watching your videos is like watching movies❤
To me this sounds a lot like Deiksra an algorithum that is used in some network routing protocols, i think also used in things like GPS, etc.
I have a question if its calculating the average time spent on each sight after an infinite amount of clicking wouldnt it rate "traps" way to high? Like if I have a website A and B that only link each other and nothing else. And then I have say 10 other nodes that have a good healthy web between them. If even 1 of them links A or B then after an infinite amount of clicking one would always eventually get stuck in the ABABABAB loop so these would both get a final rank of 0.5 and all other a rank of 0 right?
Don't forget that each website starts with a probability of (1/[the total number of websites]) which means they'll start, and remain, low probability.
Is Markov chain comes under Probability or Random Process ?
Does the limit exist only since the eigenvalue is 1?
UA-cam is now not just for entertainment. Serving as teasers for my all math courses to teach.
if you transpose the matrix then you have a markov chain. is there any mathematical significance to that?
This really a CURIOSITY feast.
i mean
BRILLIANT MAN
AMAZING VIDEO
How you make this kind of animation in video.
What you use for making this kind of informative videos.
How often did u use chegg during ur undergrad ?
Amazing breakdown. You've earned a subscriber
The girl is mine,ours !I remember Michael Jackson and Paul Mccartney singing the Girl is mine hehrhegehegejehege !
At 7:05 he said 37.5 when he should of said 0.375.
He also said "Percent" with 37.5.
Are the eigenvalues imaginary in this case?
Isn't this a Markov Chain?
Yes it is.
but I always type the url manually
Super video! I applauded for £50.00 👏👏👏👏
I've got a question: How many klick is one round of clicks?
What exactly do you mean when websites are linked?
You know, hyper-linked--the web links you click or tap on to take you to other web pages.
I wonder how much of this still holds up now that the algorithm is a million times more complicated
Also, how does this work when integrating it with a search algorithm?
great video
Why is only matrix multiplication used here?
very cool stuff
Excellent
Thank you, you have been so much help with my presentation
Hi would you make a video about information technology
How do all ranks become zero when an isolated node is added into the mix?
If a node is isolated then all entries in one of the columns of the matrix become zero (it's no longer a markov matrix). If you raise that matrix to a large power then its entries go to zero which means the ranks do as well.
Beautiful
the dating example was good, you could use it more.
you should add background music to your video
what is the font @ 3:40
This is great and amazing.
I am FAN of you
There is a way to show your appreciation - Patreon, 1 dollar per month is not that much for you, but those add up for Zach
do a video on how UA-cam recommend algorithm works.. UA-cam algorithm will recommend it to everyone
What you said about all ranks being 0 when one page is completely disconnected is not true though
Using the exact algorithm I mentioned it would be. Of course that’s not how it really works and I was just saying how the video doesn’t mention how that’s accounted for.
@@zachstar I mean if A is somehow a nilpotent matrix the ranks would go to zero. However in a general case with an isolated node that would, like shown, result in a row and column with only zeros. This doesnt have to be a nilpotent matrix. Or plays the fact that there is no self linking allowed a role such that the matrix ist nilpotent ?
@@zachstar But that's what I'm saying that using the exact algorithm not all ranks will be 0, just the rank of the disconnected node, and the ranks of the other ones will add up to less than 1 but not 0.
Dont forget the machine learning fairness AI that overrides everyone's 'biased' endorsements.
Still unsure where you got .25 from
Ye, best topic
PageRank BEST explained.
That was not a eigenvector of A.
I envy mathematicians and programmers now
at least you get to sleep
Amen yes !
I believe Google uses Panda algorithm now, named after Navneet Panda
Neo agree with you, when you talk about the matrix.
thanks for the explanation it was helpful.Infinity will be 1+1=3
What hapen To Carlos???
Am I the only one who read "RagePank" and thougt about random music ganre?
U need to make a course on brilliant. Org😍😍
First
I love math ♥
I wish I can double like this video.
You could be the next dracula
That ungrateful Bob.
Oh hey a stable distribution.....