What are Autoencoders?

Рет қаралды 121,727

Күн бұрын

Пікірлер: 37

@mdjubairpantho4142 3 ай бұрын

I have seen many videos of Martin. They are all less than 10 minutes but he made me understand the core concepts. Such a great demonstration of complexity in simpler forms .

@rachitsingh4913 8 ай бұрын

So Basically! Autoencoders are something which can generate different versions of data on which the model is trained. For Example if we train an autoencoder with certain images of Cat , than they can generate different versions of cat different than that of the original cat image like fat cat , thin cat , small cat , large cat etc.

@bharathentertainment3400 Жыл бұрын

This video is better than longer ones from other channels... simple and good explanation ✌️

@maxricardo Жыл бұрын

This is 5' video is absolutely amazing to understand Encoder/Decoder and AutoEncoders. Much simpler than many useless long videos outthere. GREAT JOB !

@HelloWorld-ji9fp 2 ай бұрын

Your videos are short and easy to understand...Thankyou sir!!

@bincybincy1 6 ай бұрын

No questions..you explained it so well..Thankyou

@alexmattyou 2 ай бұрын

Wow, A good way to remember what autoencoder is. Great video 👍👍

@tanvirhasanmonir1627 Жыл бұрын

Wow! Love the way you explain AE. Thanks a ton.

@suryavaraprasadalla8511 Ай бұрын

explained it well! why it is called unsupervised?aren't we teaching the model? how do we decide how many levels of compression is needed in encoding before it reaches latent space?

@Dron008 8 ай бұрын

Great explanation, I could understand it at last.

@TanzimMostafa-z1v 11 ай бұрын

Superbly explained!! Thanks for the video!

@Flankymanga 2 жыл бұрын

Imaging all the CSI TV series scenes where investigators are zooming (de-noising) camera pictures to get crystal clear photo shots of criminals doing their crimes. Autoencoders in real life use cases.

@CodeOmega_ Жыл бұрын

good one, thnx

@Toxic____________ 2 ай бұрын

I think now IBM and Google should start their own university cuz they explain it wayyyyyy easier than any other

@RajatVerma-y1n 4 ай бұрын

Please provide a vedio for explaining the MATHS behind the AutoEncoders.

@dam-ib9fs Жыл бұрын

Holly crap this is a master peace, perfect lesson, thank you

@theelysium1597 Ай бұрын

The video sadly doesn't explicitly state how AE are used for Anomaly detection. EDIT: The main idea is that because the AE trains on normal data, it becomes good at reconstructing that normal data (normal data -> code -> normal data representation from code normal data). However, it has not become good at reconstructing anomalous data (anomalous data -> code -> anomalous data representation from code anomalous data). Therefore, AE can detect anomalies by comparing the generated representation with the input.

@kevinkawchak 7 ай бұрын

Thank you for the discussion.

@vgreddysaragada Жыл бұрын

Very creative..well explained ..in a simple way..

@manukrishnan5159 Жыл бұрын

Deserves more likes and views ..i think the reason it got less is the accent of him which is harder to understand by most here

@anastasiya256 Жыл бұрын

where's "here"?... I think it's more likely that IBM youtube channels don't have good marketing strategy going on...

@courtneynelson4095 2 жыл бұрын

Love this. Thank you.

@keb785 2 жыл бұрын

Loved this

@IshanGarg-y1u 6 ай бұрын

rest is fine but the anaomaly detection part wasn't that clear

@sarahanjani9814 Жыл бұрын

how to determine the layer of the encoder? Is there a limit to the layers used to encode data? I really need manual calculation of this algorithm. Thanks