Variational Autoencoders

Рет қаралды 476,525

Күн бұрын

In this episode, we dive into Variational Autoencoders, a class of neural networks that can learn to compress data completely unsupervised!
VAE's are a very hot topic right now in unsupervised modelling of latent variables and provide a unique solution to the curse of dimensionality.
This video starts with a quick intro into normal autoencoders and then goes into VAE's and disentangled beta-VAE's.
I aslo touch upon related topics like learning causal, latent representations, image segmentation and the reparameterization trick!
Get ready for a pretty technical episode!
Paper references:
- Disentangled VAE's (DeepMind 2016): arxiv.org/abs/1606.05579
- Applying disentangled VAE's to RL: DARLA (DeepMind 2017): arxiv.org/abs/1707.08475
- Original VAE paper (2013): arxiv.org/abs/1312.6114
If you want to support this channel, here is my patreon link:
/ arxivinsights --- You are amazing!! ;)
If you have questions you would like to discuss with me personally, you can book a 1-on-1 video call through Pensight: pensight.com/x/xander-steenbr...

Пікірлер: 456

@abaybektursun 6 жыл бұрын

Variational Autoencoders starts at 5:40

@pouyan74 4 жыл бұрын

You just saved five minutes of my life!

@moazalomary8123 3 жыл бұрын

@@pouyan74 no the first part was necessary...

@selmanemohamed5146 3 жыл бұрын

@@moazalomary8123 you think someone would enter a video about Variational Autoencoders if he doesn't know what Autoencoders are

@moazalomary8123 3 жыл бұрын

@@selmanemohamed5146 yeah i did... 😂😂😂 and i was lucky he explained both 😎🙌😅 + the difference between them and that's the important part

@moazalomary8123 3 жыл бұрын

@Otis Rohan Interested

@atticusmreynard 6 жыл бұрын

This kind of well-articulated explanation of research is a real service to the ML community. Thanks for sharing this.

@vindieu Жыл бұрын

Except for "Gaussian" that is weirdly russian pronunced "khaussian" wat?

@arkaung 6 жыл бұрын

This guy does a real job of explaining things rather than hyping up things like "some other people".

@malharjajoo7393 4 жыл бұрын

are you referring to Siraj Raval? lol

@mubangansofu7469 2 жыл бұрын

@@malharjajoo7393 lol

@515nathaniel 4 жыл бұрын

"You cannot push gradients through a sampling node" TensorFlow: *HOLD MY BEER!*

@ambujmittal6824 4 жыл бұрын

Your way of simplifying things is truly amazing! We really need more people like you!

@debajyotisg 5 жыл бұрын

I love your channel. A perfect amount of technicality so as to not scare off beginners, and also keep the intermediates/ experts around. Brilliant.

@jingwangphysics 2 жыл бұрын

The beta-VAE seems enforcing a sparse representation. It magically picks the mostly relevant latent variables. I am glad that you mentioned ‘causal’, because that’s probably how our brain deals with high dimensional data. When resources are limited (corresponding to use large beta), the best representation turns out to be a causal model. Fascinating! Thanks

@obadajabassini3552 6 жыл бұрын

A really great talk! I have been reading about VAE a lot and this video helps me to understand it even better. Thanks!

@adityakapoor3237 2 жыл бұрын

This guy was a VAE to the VAE explanation. Really need more of such explanations with the growing literature! Thanks!

@antonalexandrov4159 2 жыл бұрын

Just found your channel and I realize how with some passion and effort you explain things better than some of my professors. Of course, you don't go into too much detail but putting together the big picture comprehensively is valuable and not everyone can do it.

@paradoxicallyexcellent5138 5 жыл бұрын

I was very interested in this topic, read the paper, watched some videos, read some blogs. This is by far the best explanation I've come across. You add a lot of value here to the original rapper's contribution. It could even be said you auto-encoded it for my consumption ;)

@hcgaron 5 жыл бұрын

I discovered your channel today and I'm hooked! Excellent work. Thank you so much for your hard work

@isaiasprestes 6 жыл бұрын

Great! No BS, strait and plain English! That`s what I want!! :) Congratulations!

@ujjalkrdutta7854 5 жыл бұрын

Really liked it. Firstly giving an intuition of the concept, its application and then to the objective function while explaining its individual terms, in a way everyone can understand, it was simply professional and elegant. Nice work and thanks!

@SamWestby 3 жыл бұрын

Three years later and this still the best VAE video I've seen. Thanks Xander!

@ejkitchen 6 жыл бұрын

Your videos are quite good. I am sure you will get an audience in no time if you continue. Thank you so much for making these videos. I like the style you use a lot and love the time format (not too short and long enough to do a good overview dive). Well done.

@ArxivInsights 6 жыл бұрын

Thank you very much for supporting me man! New video is in the making, I expect to upload it hopefully somewhere next week :)

@akshayshrivastava97 3 жыл бұрын

Finally, someone who cares their viewers actually get to understand VAEs.

@nabeelyoosuf 4 жыл бұрын

Your explanations are quite insightful and flawless. You are are a gifted explainer! Thanks for sharing them. Please keep sharing more.

@giorgiozannini5626 3 жыл бұрын

wait how did I not know of this channel. Beautiful explanation, perfectly clear. Thanks for the awesome work!

@Zahlenteufel1 Жыл бұрын

Bro this was insanely helpful! I'm writing my thesis and am missing a lot of the basics in a lot of relevant areas. Great summary!

@MonaJalal 4 жыл бұрын

hands down this was the best autoencoder and variational autoencoder tutorial I found on Web.

@ashokkannan93 5 жыл бұрын

I would like to see more videos from you. Clear explanation of concept and gentle presentation of math. Great job!

@sethagastya 4 жыл бұрын

This was an amazing video! Thanks man. Will stay tuned for more!

@ashokkannan93 5 жыл бұрын

Excellent video!! Probably the best VAE video I saw. Thanks a lot :)

@reinerwilhelms-tricarico344 4 жыл бұрын

Great! Crisply clear explanations in such a short time.

@get.ai.enabled 6 жыл бұрын

This is a LIT channel for watching alongside papers. Thanks

@venkatbuvana 5 жыл бұрын

Thanks a lot for sharing such a succinct summarization of VAEs. Very helpful!

@achakraborti 6 жыл бұрын

First video I see from this channel. Immediately subscribed!

@lisbeth04 5 жыл бұрын

I love you. I spent so long on this and couldn't understand the intuition behind it, with this video I understood immediately. Thanks

@dimitryversteele2410 6 жыл бұрын

Great video ! Very clear and understandable explanaitions of hard to understand topics.

@rylaczero3740 6 жыл бұрын

Bloody nicely explained than the Stanford people. Subscribed to the channel, I remember watching your first video on Alpha, but didn't subscribed then, I hope there will be more content on channel with same level of quality, otherwise its hard for people to stick around when the reward is sparse.

@adityamalte476 6 жыл бұрын

Really appreciate your effort of simplifying research papers for viewers.Keep it up.I want more such videos

@DanielWeikert 6 жыл бұрын

Great work. Thanks a lot! Highly appreciate your effort. Creating these videos takes time but I still hope you will continue.

@kalehermit 4 жыл бұрын

Thank you very much, this is the first time I understand the benefit of reparameterization trick.

@DistortedV12 4 жыл бұрын

This was very lucid. You are gifted at explaining things!

@davidm.johnston8994 6 жыл бұрын

Great videos man, keep them going, you're gonna find an audience!

@TheJysN 3 жыл бұрын

I hand such a hart time understanding the Reparameterization trick, now i finally got it. Thanks for the great explanation. Would love to see more Videos from you.

@agatinogiulianomirabella6590 2 жыл бұрын

Best explanation found on the internet so far. Congratulations!

@moozzzmann 5 ай бұрын

Great Video!! I just watched 4 hours worth of lectures, in which nothing really became clear to me, and while watching this video everything clicked! Will definitely be checking out your other work

@MeauxTarabein 6 жыл бұрын

Very Helpfully Arxiv! keep the good Quality videos coming

@ativjoshi1049 6 жыл бұрын

Your explanation is crisp and to the point. Thanks.

@antoinesueur9289 6 жыл бұрын

Great content ! The format and delivery is perfect, hope to see more of these videos :) . Are you planning on doing a video on Capsule Networks in the future ?

@ArxivInsights 6 жыл бұрын

More videos are definitely coming, the next one will be on novel state-of-the-art methods in Reinforcement Learning! I don't plan on making a video on Capsule Nets since there is an amazingly good video by Aurélien Géron on that topic and there's no way I can explain it any better than he did, no need to reinvent the wheel :p Here is his video: kzfaq.info/get/bejne/prZ-a8dhqpmdYGQ.html

@aryanirvaan 2 жыл бұрын

Dude what a next level genius you are! You made them so easy to be understood, and just look at the quality of the content. Damn bro!🎀

@shivamutreja6427 2 жыл бұрын

Your videos are absolute crackin for a quick revision before an interview!

@liyiyuan45 3 жыл бұрын

This is sooooooo useful for 2am and you dragged by all the math in the actually paper. Thanks man for the clear explanation!

@fktudiablo9579 3 жыл бұрын

always the best place to have a good overview before diving deeper

@kristyleist3318 4 жыл бұрын

This is great! Keep going, we need you! Don't stop making amazing videos like this

@nohandlepleasethanks 6 жыл бұрын

Great explanations. This filled two crucial gaps in my understanding of VAEs, and introduced me to beta-VAEs.

@JakubArnold 6 жыл бұрын

Great explanation on why we actually need the reparameterization trick. Everyone just skims over that and explains the part that mu+var*N(0,1) = N(mu,var), but ignores the part why you need it. Good job!

@hitarthk 6 жыл бұрын

Absolutely great stuff Arxiv Insights! Subscribed to your videos for life :)

@superaluis 6 жыл бұрын

Great channel! Keep up with this awesome project. Already subscribed and going to share this channel with my colleagues.

@seattle-bfg 6 жыл бұрын

Really really awesome channel!!! Look forward to watching more of your videos!

@SlavIvanov 6 жыл бұрын

This is great! Keep going, we need you!

@yanfengliu 6 жыл бұрын

This is really good. I like the way you explain things. Thank you for sharing!

@AjithKumar-gk7bf 5 жыл бұрын

Just found this channel ... today... one word Brilliant...!!!

@davidenders9107 6 ай бұрын

Thank you! This was comprehensive and comprehendible.

@ChocolateMilkCultLeader 2 жыл бұрын

Shared your work with my followers. Keep making amazing content

@joshbrenneman 5 жыл бұрын

Wow, love your videos. I have not worked with reinforcement learning, but I’d love to hear your analysis of other generative models.

@TheRohr 6 жыл бұрын

Great thanks for the video and the paper explanation! Really, really helpful, keep that paper explanation content!

@animeshgoyal9583 5 жыл бұрын

Simply Amazing! Thanks for sharing this. Absolutely loved it. Hope to see more videos from you :)

@rileyrfitzpatrick 4 жыл бұрын

Im always intimidated when he says it is going to be technical, but then he explains it so concisely.

@emanehab510 6 жыл бұрын

Don't stop making amazing videos like this

@famouspeople3499 3 жыл бұрын

Great video, better than many tutor lessons in university, this animation and simplified the things with simple words

@double_j3867 6 жыл бұрын

Subscribed. Very useful -- i'm an applied ML researcher (applying these techniques to real-world problems) so I need a way to quickly "scan" methods and determine what may be useful before diving in-depth. These styles of videos are exactly what I need.

@vortexZXR 4 жыл бұрын

So many ideas come to mind after watching this video. Well done!

@kanglemu6801 3 жыл бұрын

Really love this video! Good job!

@abhinavshaw9112 6 жыл бұрын

Hi, I am a Graduate Student at UMass Amherst. I really liked your video, it gave me a lot of ideas. Watching this before reading the paper would really help. Please keep it coming I'll be waiting for more.

@robinranabhat3125 6 жыл бұрын

Don't you ever stop explaining papers like this. Better than Siraj's video. Just explain the code part a bit longer. And your channel is set.

@pablonapan4698 6 жыл бұрын

exactly. show some more code please.

@shrangisoni8758 5 жыл бұрын

Yea we can't really do much until we code and see results ourselves.

@pixel7038 4 жыл бұрын

Siraj has improved his videos and provides more content. Don’t be stuck in the past ;)

@gagegolish9306 4 жыл бұрын

@@shrangisoni8758 He's explained the fundamental concepts, you can take those concepts and translate them to code. He shouldn't have to do that for you.

@dalchemistt7 4 жыл бұрын

@@pixel7038 Please stop spreading his name. He has faked his way more than enough already. Read more here: twitter.com/AndrewM_Webb/status/1183150368945049605 and here www.reddit.com/r/learnmachinelearning/comments/dheo88/siraj_raval_admits_to_the_plagiarism_claims/ And what really bugs me is not the plagiarism- that's bad and shameful in itself- but the level of stupidity this guys had shown while plagiarizing- "gates" to "doors" and "complex Hilbert space" to "complicated Hilbert space".

@maxhorowitz-gelb6092 6 жыл бұрын

Wow! Great video. Very concise and easy to understand something quite complex.

@SeanLGoldberg 6 жыл бұрын

Great episode. Came here for a good explanation of VAEs, but was blown away when you dug into Beta-VAEs and the Deepmind RL paper. Have you read the group's newest paper "SCAN" on combining Beta-VAEs with symbol representation and manipulation?

@ArxivInsights 6 жыл бұрын

Sean Goldberg Haven't had the time yet, it's somewhere in my 658 open chrome tabs though :p

@1apiano 6 жыл бұрын

I liked arxiv.org/pdf/1709.05047.pdf more, but the SCAN paper is also cool. BTW compliments for your channel, it's the only Deep Learning channel which is worth following.

@HeduAI 5 жыл бұрын

Amazing explanation to a complicated topic! Thank you so much!!!!

@satishbanka 3 жыл бұрын

Very good explanation of Variational Autoencoders! Kudos!

@user-or7ji5hv8y 4 жыл бұрын

Your explanation is so clear.

@kiwianaDJ 6 жыл бұрын

what a gem of a channel I have found her...

@sunnybeta_ 6 жыл бұрын

This video suddenly popped up today morning on my home page. Now i know my Sunday will be great. :D

@ck1847 2 жыл бұрын

Thanks, this video clarified many things from the original paper.

@abylayamanbayev8403 Жыл бұрын

Finally I understood the intuition of sampling from mu and sigma and reparameterization trick. Thanks!

@phattran4858 4 жыл бұрын

Thank you very much, I was trying to understand it, but it's much easier when I found this video!

@dmitrykalashnikov8637 5 жыл бұрын

Very useful. Great content. Continue what you're doing. Great job.

@matthewbascom 4 жыл бұрын

I like the subtle distinction you made between the disentangled variational auto-encoder versus the normal variational auto-encoder: Changing the first dimension in the latent space of the disentangled version rotates the face while leaving everything else in the image unchanged. But changing the first dimension in the normal version not only rotates the image, but changes other features as well. Thank you. Me gleaning that distinction from Higgins, et al. Beta-VAE Deepmind paper would be unlikely...

@karrikarthik6936 5 жыл бұрын

Big Fan. Would love to see videos where you breakdown some of the applications using deep learning tools.

@timurbabadjanov9115 4 жыл бұрын

That was a great explanation! Thank you so much!

@TienTaioan 6 жыл бұрын

Very clear explanation. Thank you!

@jfndfiunskj5299 6 жыл бұрын

Very clearly explained. Good job.

@ianprado1488 6 жыл бұрын

Excellent material!!

@mrdbourke 6 жыл бұрын

Epic video Xander! I learned a lot from your explanation. Now to try an implement some code!

@BDEvans 4 жыл бұрын

Brilliant explanation, thanks.

@karimabdultankian28 6 жыл бұрын

So clear, amazing 15 mins. ty,

@adamb.2390 6 жыл бұрын

That was awesome! Thanks

@betterbrained 2 жыл бұрын

You help so much with my exams, thanks man, subscribed for more high quality stuff!

@snippletrap 5 жыл бұрын

Sublime text editor is so aesthetic. Anyway, yes, great point, the input dimensionality needs to be reduced. Even the original Atari DeepMind breakthrough relied on a smaller (handcrafted) representation of the pixel data. With the disentangled variational autoencoder it may be feasible or even an improvement to deal with the full input.

@PierLim 6 жыл бұрын

Very good explanation! Thank you man!

@teegeevee42 2 ай бұрын

This is so good. Thank you!

@submagr 4 жыл бұрын

Thanks for this video. It gave a nice overall idea about variational auto-encoders.

@user-em7co7ui8y 5 жыл бұрын

Awesome explanation! Thanks!

@lordsherpaman 5 жыл бұрын

That was a great explanation, thank you!

@vladimirsafrygin7591 5 жыл бұрын

Fantastic explanation! Thanks a lot!

@bradknox 3 жыл бұрын

Great video! I have a minor correction: At 6:14, calling the cursive L a "loss" might be a misnomer, since loss is something we almost always want to minimize, and the formula of (reconstruction likelihood - KL divergence) should be maximized. In fact, the Kingma and Welling paper call that term the "(variational) lower bound on the marginal likelihood of datapoint i", not a loss.