Thank youuuu soon much! Such simple and detailed explanation, you can't imagine how much this helped me! It gives me the idea of 'recommendation' and their matrices Thanks again ..
Can I use the matrix factorization technique for recommendation where the scenario is like I have 972 unique user and 3810 unique items and 24 unique country id. Like in a common movie recommendation ratings are there and we predict ratings and then show recommendation system. In my case I have country id instead of ratings. is it fundamentally incorrect or I can go ahead with this?
Great examples, but I really wonder. Doesn't this kind of peration take too long to run? I mean running this kind of algorithm would probably take 30ms, but considering you might have thousends of users trying to run the algorithm simultaneously, it might be a pain in the ass. Aren't there better ways to deal with it?
Well what about the Google search engine? What kind of recommender system do they use? How do they decide the precedence of the search results for an user? Do they use content based filtering based on the search keywords, or does they use a hybrid recommender system where they first collect sites from the internet based on the search terms and then use collaborative filtering based upon all the users who have searched using the specific terms and then predict which link the user is most likely to click on based on what others have clicked on?
i wold like to know what is the filed for "software engineer" to study in order to control this filed and specializes in this field ? thanks (i'm looking to hire someone at my location)
There must be a better way to do this...I don't identify with most of what the "big sites" recommend me. UA-cam, for example, keeps recommending me chess related videos when I've never watched anything remotely chess related as far as I know. I guess this must mean that my tastes are somehow adjacent to the subject of chess, but it sounds like the system would yield better results overall if it could somehow realize that I am not responding to that particular subject.
+Erik Pontifex your concern is genuine but another aspect of machine learning (which is basically a field of AI) is to try and predict things you might like just like netflix . one way of doing this is grouping people (clustering ) with similar tastes together . This might be what youtube does and it just might happen that the cluster you are in consists of people who like chess thus the recommendations.
A very fascinating lesson in how the recommendation system of Netflix, Facebook works, etc. Thank you so much CS50 staff!
WHAT A SIMPLE AND PRECISE EXPLANATION....... thank YOU SIR
ITS SOOOOOOOOOOOOOOOO HARD
Great video. I appreciate how you have explained the concept in a much comprehensible way.
When i audited this course 7 years ago, i had no idea i will do ML for a living.
Thank You.Explaination is so simple so anybody can easily understand.
Thank youuuu soon much!
Such simple and detailed explanation, you can't imagine how much this helped me!
It gives me the idea of 'recommendation' and their matrices
Thanks again ..
I thought your Linuz Torvalds, but thanks, great explanation!
same, haha, I came across thinking it was Linus talking
You are!?
Ya exactly..... 😀
quite true ... thought it was him!!
maybe he's a fan..or he's an impersonator of Linus hehehe
Amazing lecture. Thank you.
Thanks for the simple explanation.
this was 8 years ago, amazing
Thanks for simple and brief explanation.
Thanks in a million. Awesome. Where have you been all these years.
thanks for your explanation about the remcomendation 's system sir
Great video! Very explanatory and easy to understand!
Thank you. It's very simple and easy to understand
thanks for you, and for using simple example for understanding this topic
Linus Torvalds explaining us recommendation systems... interesting
Can I use the matrix factorization technique for recommendation where the scenario is like I have 972 unique user and 3810 unique items and 24 unique country id. Like in a common movie recommendation ratings are there and we predict ratings and then show recommendation system. In my case I have country id instead of ratings. is it fundamentally incorrect or I can go ahead with this?
omg, it is an amazing lecture!
thanks for your great and simple explanations
Where is the rest of this course ? I dont see the playlist
Unfortunately Netflix seem to have throw all that work away
Great presentation.
A fantastic way of explaining the content. I understood the underlying concept. Thank You so much.
Your not welcome :(
Great explanation sir
Thanks
Really helpful
Thanks sir good work👍
Didn't know Linus Torvalds taught CS50
IF YOU DIDNT KNOW THAT THEN YOU CANT BE IN CS50 I AM IN THEIR UNIVERSITY FREE SCHOLORSHIP
it seems like an efficient tag system is more important than recommender algorithm
Great examples, but I really wonder. Doesn't this kind of peration take too long to run? I mean running this kind of algorithm would probably take 30ms, but considering you might have thousends of users trying to run the algorithm simultaneously, it might be a pain in the ass. Aren't there better ways to deal with it?
very nice explanation
omg I just saw this video on 2020
thanks, very clear explanation!
good job...thanks for explaining in simple way :)
amazing lesson. thank you professor
clear explanation. thank's
Excellent explanation! Shukran.
EYVALLAH BEY
Thanks for your lectures. Is it code Python?
For a moment I thought he was Linus Torvalds
진짜 굉장한 강의들이다 정말 하버드 진짜 만세다 ㅠㅠ
Brilliant explanation!
Thanks for sharing this video.
Great video! Congrats
How do you determine the value of k in SVD? Is stochastic gradient descent used for that or is it a completely different method?
Well what about the Google search engine? What kind of recommender system do they use? How do they decide the precedence of the search results for an user? Do they use content based filtering based on the search keywords, or does they use a hybrid recommender system where they first collect sites from the internet based on the search terms and then use collaborative filtering based upon all the users who have searched using the specific terms and then predict which link the user is most likely to click on based on what others have clicked on?
This is an another topic, called Information Retrieval
Google uses Page Rank algorithm for listing the search results and I think that they use hybrid system for recommending similar searches.
Thank the Math Gods that sent you to me!
Great explanation!
> Want to learn about recommender system
> Sees Bechdel Test
> Pauses Video
> Search "Why is Bechdel Test even necessary?" For 30 minutes
Kevin Eontrainer lol
very excellent talk.
Just awesome!
THIS man is an og explainer lol
i wold like to know what is the filed for "software engineer" to study in order to control this filed and specializes in this field ? thanks (i'm looking to hire someone at my location)
Scaz for president!!!
you look like young Sheldon's father
nicely done, i will implement content based since it is more reasonable in early
thanks a million for you
lol the CC spells boolean as "bullion"
thanks, you helped me in my GP
Yo after those years this lecture is so good 😱😱
is it possible to apply recommender systems to Intrusion Detection Algorithms?
no its not possible
anything is possible if you are brave enough.
Hi sir, can you please give me a lecture on Building Recommendation System ?
nice videos
Thank you .
Thank you sir.
thank u so much
There must be a better way to do this...I don't identify with most of what the "big sites" recommend me. UA-cam, for example, keeps recommending me chess related videos when I've never watched anything remotely chess related as far as I know. I guess this must mean that my tastes are somehow adjacent to the subject of chess, but it sounds like the system would yield better results overall if it could somehow realize that I am not responding to that particular subject.
+Erik Pontifex your concern is genuine but another aspect of machine learning (which is basically a field of AI) is to try and predict things you might like just like netflix . one way of doing this is grouping people (clustering ) with similar tastes together . This might be what youtube does and it just might happen that the cluster you are in consists of people who like chess thus the recommendations.
Nice
thank you
thanks,for video
amazing, thanks
Thank you a lot!
Lol i saw this on my reccomandation
Thank you
Thanks Sir
Hi Sir
Will you provide the sample code of the video about?
Thanks
what is the cross domain recommendations please explain it
please give me some idea about cros domain recommendations
it helps. thanks
i came from clevered.com pre class video
Really nice i also came
@zeyrox you gotta be kidding me nah?
no i am not lol
ohhh ok KOKO MO MUJEH BI DO
SAME BOI
:D
Jack Vu is here.
Next week i have my exam based on recommender systems, hope i will pass...