Build an Autoencoder in 5 Min - Fresh Machine Learning #5

This video is all about autoencoders! I start off explaining what an autoencoder is and how it works. Then I talk about some use cases for autoencoders and the special types of autoencoders we use for each of them. Finally, I programmatically go through an example of a simple autoencoder, followed by a demo.

The code for this video is here:

I created a Slack channel for us, sign up here:

Autoencoder live demo in the browser:

and here are some great links to read up on autoencoders:

I love you guys! Thanks for watching my videos, I do it for you. I left my awesome job at Twilio and I'm doing this full time now.

I recently created a Patreon page. If you like my videos, feel free to help support my effort here!:

Much more to come so please subscribe, like, and comment.
Follow me:
Signup for my newsletter for exciting updates in the field of AI:

[TimmyBlumberg]

Finally. A YouTuber that is actually unafraid to get into the technical details of machine learning! Thanks for making your great videos!

[medmed]

Thanks Siraj for great video. Do you have any plan for another video on time series (sequence) autoencoder?

[bosepukur]

thanks siraj for giving me a place to start learning

[nicholascantrell]

Just woke up and this video was a nice surprise. Thank you for making it!

[akrylic_]

I'm blown away by every video you do. You've inspired me to do a lot of this kind of stuff on my own. Keep up the great work!

[robertotomas]

If I could up vote this twice I would. This was amazing

[petermitrano]

another excellent video. Love the enthusiasm
PS: found you via HH

[brunoacosta]

i get myh mind blown away every time i watch one of your videos.
thanx a bunch for that

[johnhammer]

Dude do you ever sleep ?. So much work in making videos and keeping up with the current papers. Great. Thanks so much dude.

[invinity]

best of the best!!!! love your videos, and the content you are choosing is cutting edge! shows your passion and commitment!

[gabrielbagur]

Great! That bit slower helps a lot to follow up, with no penalty for fun dynamism. I appreciate that there is no long and boring overhead, so normal in others youtube videos.

[Happylife-zvmf]

Wow learnt few things.. Thank u Siraj!

[SussyBacca]

Holy shit. Magical video. You don't waste any time. I love you. You're also insanely accurate and funny too. Achievement Unlocked: Perfection

[Yan-osjl]

What to write in import input_data?? Do we have to write file location?

[vinayak]

Hi Siraj! Thank you for all of your awesome videos! You've really helped me come far in the field of Machine Learning! I do have one request though, could you make a video that explains the mechanism of designing an autoencoder in tensorflow slower? I have an ongoing major project that involves autoencoders and have never actually used them before. If not, could you please direct me to a great source in your opinion that would do the same? Thank you so much, and I apologize for any hassle my request may cause!

[KTbK]

Sirajology, I love you! =D

You found the best format to represent ideas and approaches!
I am a little bit afraid, that you will cover only few hot topics and stop. Please, reassure me, that you will continue forever. Into the depths and wild. 😊

[ateoc]

I have a question : on 4:34, line 33 and line 40, why reduce the sum of square of X-Z? Z had sigmoid already, so I think it can`t be 'X = Z' . or, had MNIST data transformed?

[badashphilosophy]

The people who come up with this stuff are legends, its like computational philosophy that actually has practical value😂. U need a suggestions blockchain voting system for your school where people can philosophise new coding technologies and the as the idea grows and takes off the creators token value increases raise in value. People can collab to smash out the ones worth doing and everyone can contribute in some way. Ud be pumping out awesomeness at an unprecedented rate.

[againstballveer]

Hey Siraj, where did you learn all this?

[Aphsun]

can I use autoencoder for feature enhancement in text summarizing?