You're the greatest UA-camr I've ever come across since I started my ML and DNN self-taught journey. Please keep on creating more of this top notch content. You have a person here who leaves this video with a huge smile on the face because everything was crystal clear. Greetings from a PhD student in Finland!
Thank you. The pictures you printed and sketched were awesome. Appreciate the simplicity of this complex topic. Wish there was a full video of Handling Missing Data with complete intuition because we can't be experts of all fields and thus deciding whether to drop, impute etc a column is right or wrong on what situation. I liked the new column option in your short video so the ML model can at least get the information for what we're dropping. Same goes for Outlier detection and removal techniques.
I just started learning about Neural Networks and today you showed me a different way to use Neural Networks as Auto Encoders. I like the denoising images one, cuz it's different from other (shark) classification problem. The way you teach about a topic is really amazing sir. 💛
Great video, I was trying to understand auto encoders for a long time, all the other videos had just the keywords but your scenario based explanation made it clear and understandable. Thank you
What would you recommend, can we use autoencoders for feature extraction from the text data for sentiment analysis? Would it work? I mean the idea overall , is it suitable or is it just a bad idea
Amazing work! I appreciate the amout of work being put to beak down the concepts! I wish If we can have a video about EDA, that'll be amazing as well. Thanks!
@@underfitted As an introduction to autoencoders, this has definitely got me excited! Can't wait to get hands on with this. Use cases are insane! It's relevance to human-level pedagogy is also fascinating.
It seems it would be more efficient for an auto encoder to look for at least 500 key patterns in an image.. i.e. deep, shallow, dark, light, school of fish, ect ect.. Need like a contextual autoencoder.. then the ML can answer questions better and need much less training.. we dont want to overfit data, And we dont want to train with allot of data.. we also want ML to learn on the fly.. We need it to be able to tell the difference between friend or foe.. fake or real.. unless it can do this, its very limited.. You can solve fakes by looking for evidence of authenticity and cross checking across many domains.. i.e. you dont feed a horse petrol..
One of the best and most intuitive explanations of what an autoencoder is. You are a great teacher!
Wow, thanks!
best introduction to Autoencoders anywhere on the internet. 💯
You're the greatest UA-camr I've ever come across since I started my ML and DNN self-taught journey. Please keep on creating more of this top notch content. You have a person here who leaves this video with a huge smile on the face because everything was crystal clear. Greetings from a PhD student in Finland!
This was amazing! Even I am a huge fan of Autoencoders it is one of the most amazing thing in Deep Learning.
Thank you. The pictures you printed and sketched were awesome. Appreciate the simplicity of this complex topic.
Wish there was a full video of Handling Missing Data with complete intuition because we can't be experts of all fields and thus deciding whether to drop, impute etc a column is right or wrong on what situation. I liked the new column option in your short video so the ML model can at least get the information for what we're dropping.
Same goes for Outlier detection and removal techniques.
Greatest thing ever!
Really clear and accessible description of Autoencoders. Thank you for putting all the time and effort into this. Looking forward to more.
You're very welcome! More is coming!
I just started learning about Neural Networks and today you showed me a different way to use Neural Networks as Auto Encoders.
I like the denoising images one, cuz it's different from other (shark) classification problem.
The way you teach about a topic is really amazing sir. 💛
Glad I could help!
What a teacher!!! Amazing!
Thanks!
Great abstraction of how autoencoders (as a wonderful idea) works. Thank you, and please carry on what you are doing.
This is my first time coming across using autoencoders for anomaly detection. Super clever!
Thanks for sharing (and recording this twice 😂)
Great video, I was trying to understand auto encoders for a long time, all the other videos had just the keywords but your scenario based explanation made it clear and understandable. Thank you
This is a great explanation. Simple, clear, not too long.
Glad you think so!
This is such a nice explanation of autoencoders. Thank you!
You're very welcome!
This video is interesting, easy-to-follow, and informative. Than you.
Thanks!
Great Explanation Ever !!!! . Thank you Sir..
I've built an AGI which is very loosely inspired by how an autoencoder functions. It's an abstraction of the essence of this same idea.
OMG! this is amazing! going to be tacking this onto the front of some of my personal NNs! thank you for the inspirations!
You are so welcome!
Your videos are insanly Good.
Thanks
You're amazing, im studying AutoEncoders
Thanks so much!
Thanks, Santiago for putting so much effort into your videos.
Glad you like them! I don't, so I'm trying to improve. So many things I want to do better!
This is Amazing; such an intuitive explanation. Thank you
You're very welcome
Great! Following on Twitter and now here!
Thanks, Jairo!
Very entertaining! brilliant explanation
Glad you liked it!
Thank you so much for this video, really great example, i just learnt about autoencoders today, i would be looking forward to more of your uploads
Glad it was helpful! More is coming!
Sir, you are just great. I want to ask whether this approach would be helpful to auto-label the images that contain different objects. Thank you!
Thanks, Great explanation
You are welcome!
What would you recommend, can we use autoencoders for feature extraction from the text data for sentiment analysis? Would it work? I mean the idea overall , is it suitable or is it just a bad idea
Amazing work! I appreciate the amout of work being put to beak down the concepts! I wish If we can have a video about EDA, that'll be amazing as well. Thanks!
Great suggestion!
Amazing video!
I am now thinking of using Autoencoders on anomaly detection on sensor values of IoT systems. Do you think that will work?
I don't see why not.
When will the new video of the autoencoders without bottleneck be released, please?
You are greaaaaat man!!!!
brilliant!
Thanks
Awesome !
Thanks!
Good
At the 0:12, I though you would say "I used Auto Encoder to....."
Thanks!
@@underfitted As an introduction to autoencoders, this has definitely got me excited! Can't wait to get hands on with this. Use cases are insane! It's relevance to human-level pedagogy is also fascinating.
You should change the name of the channel from Underfitted to Underrated for accuracy.
Lol. Thanks!
Came for the sharks, stayed for the autoencoders
Thanks, Esra!
Ooh, so you're the guy behind bnomial
That would be me, and a friend.
Wait, no picture of a sharkyoctopus??? I'm disappointed!!!
I tried… I promise 🫣
It seems it would be more efficient for an auto encoder to look for at least 500 key patterns in an image.. i.e. deep, shallow, dark, light, school of fish, ect ect.. Need like a contextual autoencoder.. then the ML can answer questions better and need much less training.. we dont want to overfit data, And we dont want to train with allot of data.. we also want ML to learn on the fly.. We need it to be able to tell the difference between friend or foe.. fake or real.. unless it can do this, its very limited.. You can solve fakes by looking for evidence of authenticity and cross checking across many domains.. i.e. you dont feed a horse petrol..
Cacetada, senti até minha mente expandindo ao infinito
Lol
Help, I don't know why I'm here or why I watched the whole thing, i need an adult
Lol
You should've used an AI model to de-blur the original video :D
Lol