Node feature

[shrimonmuke0202]

I want to calculate the average of node features for a graph. How can I do this?

[rusty1s]

What do you mean with average? If you want to find the average value for each feature across the node dimension, then you can simply run:

data.x.mean(dim=0)

[shrimonmuke0202]

I want to use dgl.mean_nodes method in PyG.

[rusty1s]

I see. This is equivalent to global_mean_pool

x = global_mean_pool(data.x, data.batch)

in a mini-batch scenario.

[shrimonmuke0202]

I want h^{'} = \frac{1}{|v|}\sum_{i=1}^{g}(h_{i}) where h is the feature of each node and |v| is the number of nodes in a graph g.

[rusty1s]

Yes, that's exactly what data.x.mean(dim=0) (on a single graph) or x = global_mean_pool(...) in a mini-batch of graphs is doing.

[shrimonmuke0202]

Thanks for your reply.

…

On Tue, Nov 2, 2021 at 1:51 PM Matthias Fey ***@***.***> wrote: Yes, that's exactly what data.x.mean(dim=0) (on a single graph) or x = global_mean_pool(...) in a mini-batch of graphs is doing. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#3420 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AQ7MTCRX7MQXDXTOOVMJ2Q3UJ6NR7ANCNFSM5HFSHBCA> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>.