วีดีโอ

Thanks a lot .. where can I find the code

The audio is unfortunately not the easiest to follow. Would be great if we got corresponding English subtitles. I can only see Italian auto generated subtitles

Great explanation, thank you!

Thanks for the efforts but who the heck is typing unmute?

GAT Damn these tutorials are good!

sorry,I can't English, Can u Chinese?

That was so nicely explained, thank you ❣

Great tutorial, but he speaks too fast and scrolls through codes too much. It is hard for me to concentrate.

There is lot of background noise, which is really distracting.

Can a nodetype .y be configured on more than one node type, so it would be possible to predict targets for multiple node types?

Great tutorial - I'm working with chemical datasets trying to predict reaction outcomes with GCNNs and this tutorial is a lifesaver. It's pretty incredible how many resources there are out there for machine learning and the fact that something like pytorch geometric is available to everyone for free is just awesome. Thank you for the good work sir.

15:02 what about say, batch=[198], ptr=[12] second last object printed with for loop? What do these batch=[198], ptr=[12] actually means?

@94longa2112 Are you able to upload notebook 4? The others exist but not this one. The community would appreciate it.

good video

I am not sure, but I think there is a mistake on the GraphSage slide (24:43 into the video): AGG({h-subscript u-superscript k-1...}). Shouldn't it be: AGG({h-subscript u-superscript k...}) ? At least, this is what I read from the original paper. If not: Could you explain?

Thank you

Thank you

You guys saved my as* <3 Such precious knowledge, thank you!

great tutorial!!!!

In computation graph of B the branches of node E are not correct at 15:50

Much appreciation for this tutorial series!

good video，know many details about how to training GAT thk u

Great tutorials Antonio. If your dataset is an edge-list with edge-level features and no node level features (e.g., no x feature matrix), can this still work with this autoencoder?

Hi, i would like to know if it's possible to assign a different number of clusters to each input graphs according to its size? I have graphs ranging from 10 nodes to 150 nodes.

Great stuff very valuable

how to create a graph for custom dataset that will work for autoencoders

On minute 10, what the parameter 'ptr' represents?

Hello, great video. Is it possible to make a graph classification RecGNN?

Hey Antonio, Amazing videos and thanks for this! However, in this tutorial, as pointed out by @Arjay2186, the model is just learning to predict a lot of non-existant edges and the accuracy seems high; F1 score would be a better metric in this case. For anyone following these tutorials I suggest you compute pos_logits and neg_logits separately and then concatenate them, also add a margin (~1.5) to neg_logits, the model should learn better features. You can also try to make the model more complex (more convolutions, bigger feature dimension, etc.). This will bring down the number from 3M+ to a comparable magnitude. I tried this and it worked.

Dita-set :D

Graph sage also does sampling of neighbours, that's the inductive part of the model. Can you comment on that, it is missing in your current lecture

Although we have label only available for author node, can we also make label prediction for paper nodes?

-9e15 is a very large negative number

How computationally expensive are these kinds of networks compared to something like a more traditional deep learning architecture such as CNNs?

13:15 what is the purpose node_attrs.index += 1 similarly others edge_index.index, graph_idx.index, etc.

at 16:39 if g_w = diag(Uw) is filter v*w = U^H g_w Uw ...... (1) why Uw in RHS of (1) why not Uv? Anyone please clarify me. Also what is H in (U^H)

Amazing tutorial ! Thanks =]

Amazing!! Super didactic, thank you =]

Thanks for the tutorial, I don't get one thing on the test of VGAE, why do we encode on the training and not on the test set?

Thanks so much for the series of tutorial. I started to feel lost in this episode 😢

This tutorial is such a life saver, because its pretty dry out there for gnn stuff.

In pytorch geometric, when should I use a tuple as the in_channel? what exactly is the source and destination dimension in in_channels that the pytorch documents refer to?

How do you create an undirected bipartite graph?

Beautiful lecture and tutorial, In Slide 59, is it q(z\x) we do not know or p(z|x). Can you confirm

Hey Antonio insightful lecture,can we somehow model arc42 chollet challenge as gnn problem?

Hello, could you show me how you make recommendation on new users?

This tutorial doesn't do latent space visualization, very important aspect of gauging if VAEs are being trained correctly. Could you share a video which does this too?

Great Tutorial！

My graphs was created from images and saved as pt file how can i apply graph mining later please with this format

Hi nice video. I am very new in GNN. I am having a problem running the code. It is giving me the following error: running_mean should contain 150 elements not 64. Can you please help me with that.