MIT 6.S191 (2022): Convolutional Neural Networks

MIT Introduction to Deep Learning 6.S191: Lecture 3
Convolutional Neural Networks for Computer Vision
Lecturer: Alexander Amini
January 2022
For all lectures, slides, and lab materials: introtodeeplearning.com​
Lecture Outline - coming soon!
Subscribe to stay up to date with new deep learning lectures at MIT, or follow us @MITDeepLearning on Twitter and Instagram to stay fully-connected!!

[@luis-azanza]

I love the explanation of the actual architecture and the thought process behind the elements. Thank you so much for sharing. I already have my code up and running where I'm playing with this exact example :)

[@tahoon2009]

Beautiful and smooth explanation of the convolutional filter in action, first time to hear it and went through right away

[@ajaytaneja111]

This is such a powerful lecture in the Deep Learning Series. Each time you watch it or in every release of the series each year, it leaves a mark in your thought process and one feels intellectually enriched . Such complex concepts and so simplistically explained! Thank-you Alexander Amini. #MIT 6S.191 rules the world with regards to the Deep Learning fundamentals content.

[@281019641]

Thank you for the lecture. so clear and concise, the thing I love most about these lectures is explaining why we are doing something, and a taste of what other applications are there.

[@andreas.karatzas]

I really liked that you shared one of the research fields that you are active (autonomous driving), which I'm sure motivated most of your audience even more on the subject.

[@_NguyenTheTrung-fs1ot]

Many thanks for breaking down complex subject matter into easily graspable blocks !!!

[@shubham-pp4cw]

this is amazing as lots of CNNs contents and even higher level content explained in just 50 minutes. So beautifully and nicely content were explained easy to understand, helping me building strong fundamentals in AI. and also allowing me to deep dive in Deep Learning. Thnaks a lot for such a wonderful content. And Keep uploading the videos in Deep Learning

[@vantagepointmoon]

23:23 -> mind=blown when it all ties together in a very intuitive way

[@astrojaswanth4986]

Really love the way u explain it.. It really helps people all around the world to learn about neatal network and Deep learning. Thank you sir and mit

[@FilippoGronchi]

Excellent and clear as usual Alexander...thanks a lot

[@chai8387]

easy to understand & concise great stuff

[@henryjavier4797]

Excellent presentation!!... Thanks...

[@doctorrobot9757]

Thanks for the great effort. I hope the slides would be available soon. The links for the slides are broken for CNN and Deep Generative Modelling

[@Amanpreetkaur16]

Just love you for everything thank you so much 😍💕

[@paragon9671]

What a great presentation, you explained the whole concept in such a way that my grandmother could understand. A big thanks, I cannot thank you enough. Please kindly upload the slides "Deep Computer Vision". Thank you once again.

[@citizen240]

“… my grandmother could understand.” PROPS to your grandmother ‼️

[@mehdismaeili3743]

very amazing video.thanks.

[@islamnuralam4402]

Thanks you for your explain so brief

[@priyacharaya7155]

Awesome explanation

[@thaison7588]

Very interesting!!

[@ShaidaMuhammad]

Thanks Alex ...

[@ahmedb2559]

Thank you !

[@kyrgyzsanjar]

Awesome!

[@amir4207]

perfect

[@harinib4353]

Is thers a transcript available in pdf for this video? Or lecture notes or something?

[@style24_7]

Waiting 👀

[@Generic12321]

6.S191? More like “Success” 191 🎉🙌🏻

[@awaisahmadsiddiqi6505]

can someone please explain at @26:22 , how 3 feature maps are formed in convolution when there is only one input image?

[@AAmini]

Depending on how many convolutional filters are in your layer you can learn an arbitrary number of feature maps. So in this layer we hav learned three convolutional filters, each filter outputs one feature map, thus we have a total of three output feature maps. Recall that the convolution operation typically takes as input at least two arguments (1) the filter size and (2) the number of filters; the number of outputs and size of the outputs is directly dependent on these two arguments.

[@awaisahmadsiddiqi6505]

@@AAmini Now its clear. Thank you so much. :)

[@sarmadf.ismael2155]

the Lecture note link not working

[@felipevaldes7679]

there is a missing paren at minute 46

[@gantabalaraju7287]

I need slides for this lecture . Can anyone help me

[@aLiGeNiUs.TheHackers]

AI Kingdom kzbin.info

[@TomaszGradowski]

Great lecture. But please, let Lena retire from computer science.

[@stuckbug3586]

Who's Lena?

[@bartekbinda6978]

@@stuckbug3586 en.m.wikipedia.org/wiki/Lenna

[@mishrr]

Also waiting to get likes from professor

[@priyacharaya7155]

Awesome explanation