Introduction to graph neural networks (made easy!)

[@bnjmn7779]

Hope there will be more videos about GNNs, would love to see a variety of real problems being represented by Graphs and solved using GNNs!

[@prachi07kgp]

Very nice video with such easily understandable explanation of such complex concepts, thank you. I don't know why you stopped making videos, you are good at them, I watched all four of your videos

[@ScienceMasterHK]

Incredible explanation! You made it simple to the point.

[@newbie8051]

Amazing explanation, was reading about GNN's for an intern assignment Thanks !

[@ShikhaMallick]

Thanks for the cool explanation! Subscribed👍🏻

[@mikeCavalle]

my first impression is "what a nice background ambiance. thanks for the topic and level.

[@jackkensik7002]

Great video, well made👍

[@TheVishnu3333]

Thanks for putting the effort into this well made video! Was very helpful

[@draziraphale]

Thanks for this, I'm using this to help a student with a GNN journal publication.

[@thousandTabs]

Awesome video and explanations! So many good resources to check out, thank you for making this! Looking forward to more

[@Lily-wp5do]

Simplest explanations so far

[@ananthakrishnank3208]

Nice. Your website is cool as well!

[@amanrv]

Pretty great! I would love to see some example applications of these models in a future video. Subscribed!

[@MahimDashoraHackR]

Wonderful video jacob , Please also make a video on implementing GNNS ,GATs on Pytorch /TensorFlow and explain how math works in code .It would be really helpful

[@Hanyao8]

Brilliant, very helpful!

[@jerryyang7011]

Dots connected - great work.

[@saurabhmahindre]

Amazing overview!

[@christiansinger2497]

Great content!! Greetings from Germany

[@xphn1985]

Great content!! Thank you!

[@Septumsempra8818]

WE WANT MORE!!! 1yr is too long a wait

[@milandoshi7640]

Thanks. Can you make a practical video on how to generate a GNN.

[@pennyfarthingchapel]

Surely k is not the number of hops? From the paper it says "We use superscripts to distinguish the embeddings and functions at different iterations of message passing."

[@mrigankanath7337]

at 3:41 shouldn't it be "l" instead of "k"? I mean it should be a layer, not hop (the previous layer will affect the next layer)

[@jacobheglund4245]

I agree, the embedding computed by the previous layer would be used to compute the embedding of the next layer. I use hops and layers interchangeably here because the embedding for node i (h^1_i) computed by the first layer of the GNN will only consider information from the 1st hop neighborhood of node i. Then in the 2nd layer, the GNN will again propagate information from the 1st hop neighborhood of node i, but now the embeddings of nodes in the 1st hop neighborhood contain information about nodes in the 2nd hop neighborhood.

[@mrigankanath7337]

@@jacobheglund4245 Yes I get it what you arr trying to convey . Thanks,and really cool videos

[@sunaxes]

Sorry to me GNN are “just” being selective about how to connect one layer to the next through the adjacency matrix. We simplify the basic dense layer system which then enables fast convergence of the network… right? Conv nets do the same. They establish a neighborhood to each input. So if previous approaches do not leverage this inherent structure in the data, they cant do as well. Let me know if I m missing something.

[@thecutestcat897]

thanks a lot!

[@vijayrameshkumar8522]

Great content! Hope to see more videos.. Can you make videos with real world examples.. with code..

[@ThankYouESM]

How is GNN better than the BoW algorithm?

[@cybervigilante]

So I clicked on the link to the first video on graph neural networks and it took me right back to the Second video. Bad graph 😃

[@arshsharma8627]

i bet its a great video but its too technical, can someone recommend something more basic

[@TheGroundskeeper]

This is not the area to begin machine learning

[@WahranRai]

May be reduce the speed of your speech !

[@johnspivack6520]

Sorry, this is a bad video. I gave a thumbs down. It's phony, basically because you don't discuss the key concepts at all. You show some fancy formulas but also never get to a good conceptual explanation. Please do better or don't take our time. No offense.