How does YouTube recommend videos? - AI EXPLAINED!

I did not understand how the nearest neighbor is coming into the picture. Is it just selecting the top K from the softmax ? Also, how is the softmax output the vector for the video.

the algorithm is horrendous. the predictions are retarded.

It's so simple, if there is anything wrong On yT, blame it on the algorithm.......... Insane

Love the in depth explanation and not just discussing the basic stuff on every Recsys intro video out there. Thanks!

Amazing Explanation. Also Thanks for sharing the relevant resources :)

was not expecting this to get so deep, was happily surprised

Thanks a alot this video save my life for tommorow presentation! ❤🎉😊

Collaborative Filtering, Matrix Factorization and Deep Learning is explained well. It would be nice to put with its notebook to understand how it works?

This channel should have millions of viewers.

It's actually not svd algorithm, it's just svd-like algorithm, you can remove the middle matrix, just keep other two matrix is enough

This is sn AI Viodeo and chanell FK this trap

Bro is this math class or what

Very helpful

im here because of a spike in one of my uploads

next time make a thumbnail where you open your eyes even bigger and maybe you will get more clicks.

It remembers what you watched and recommends the same video over and over

I watched the video. I've been recommended this video twice after the fact. Maybe the algorithm predicts me speaking up and commenting?

this MATTrix. Bro, its not MATTrix, its MATErix

Please look up how to pronounce "matrix," "lenient," and "cos." For a guy without an accent, your English is terrible.

A grain of sand.
Talking schitt to: the SAHARA
...will DROWN in a POOL
- of "his" NIGHT/mare 'n (error)
..."*sauce" 👈💩🍅💖💖💖💖




*(saw/us) 🤺

For Item-Item collaboration, is it "videos based on similar videos users liked"? or "videos based on similar videos similar users liked?" They can't correlate random likes can they?

Oh damn, I didn't even know likes did anything, do you have a source for this infomation?

+1

mAtrix

Wow i learn a lot in this High Quality video!,
liked and subscribed!
Also i dislike a lot of others videos in order to improve my vote in this one!

Thank you for sharing this video sir

so i thought when item features are used than it's not CF but Hbrid beetween content based and collaborative filtering

Thank you so much

The irony is I searched this. It wasn't recommended to me.

Do a video about stocks man your smart asf I’m a rapper/nerd living in the hood I love these kind of videos. I love learning things like this. I’m happy why I stubbled upon your channel.

this video is not easy to understand, unless you know probabilities and math. You need use simple language and way to make ppl understand. Sorry not very helpful for many

This was amazing, thank you

Sir Which Software do You Used to Edit video...??

There's the yt spam filter , and timed out in comments , and lag but that's prob in the formula too for the recommends end factor .

I like the way you explained this simply!

I have been watching your videos since long time and your mathematical approach is very unique and easy to understand. This channel explains ML better than top Universities. I follow Bishop's Pattern Recognition Textbook and all your mathematics are based on his textbook which is valuable than Gold. You are making the problems of hardest textbook for ML easy! Thank you for doing such a rich and quality work. I'm sure in few years this channel will start using it's own Collaborative Filtering ;)