An Introduction to Graph Neural Networks: Models and Applications

  Рет қаралды 286,593

Microsoft Research

Microsoft Research

Күн бұрын

Пікірлер: 124
@leixun
@leixun 3 жыл бұрын
*My takeaways:* 1. Background 0:48 2. Graph neural networks (GNN) and neural message passing 6:35 - Gated GNN 26:35 - Graph convolutional networks 29:27 3. Expressing GGNNs as matrix operations 33:36 4. GNN application examples 41:25 5. Other models as special cases of GNNs 47:53 6. ML in practice 49:28
@shawnz9833
@shawnz9833 3 жыл бұрын
can you help with that: what is a MLP in "you multiply it with a single layer MLP" @23:29?
@leixun
@leixun 3 жыл бұрын
@@shawnz9833 Multilayer perceptron
@shawnz9833
@shawnz9833 3 жыл бұрын
@@leixun Cool, thank you mate!
@leixun
@leixun 3 жыл бұрын
@@shawnz9833 You are welcome mate, I am doing some deep learning research as well. You are welcome to check out our research talks on my KZbin channel.
@syllogismo
@syllogismo 3 жыл бұрын
Don't know why people are criticizing this video and the audience. Great introduction to graph neural networks!
@susmitislam1910
@susmitislam1910 3 жыл бұрын
So GNNs are basically something like calculating word embeddings in NLP. We have a dataset describing the relationships between pairs of words (nodes), and we want a vector representation that reflects how often they co-occur (weight of the edge between the nodes), i.e., how much relatedness the two words have. Once we have such vectors, we can build a vanilla, recurrent, or convolutional neural net to find out a mapping between the vectors and the output we desire.
@iltseng
@iltseng 4 жыл бұрын
At time 34:11, the (dot product of matrix A and matrix N) should be [ b + c ; c ; 0 ]
@MingshanJia
@MingshanJia 4 жыл бұрын
Actually, here what we use is the incoming edges (see 14:55), but that is true the slide is confusing about that.
@michaelkovaliov8877
@michaelkovaliov8877 4 жыл бұрын
It seems that the mistake is in the graph adjacency matrix, because the result vector is true given the drawing of the graph.
@PenguinMaths
@PenguinMaths 4 жыл бұрын
That is a mistake in the slide, A should be transposed to describe the incoming edges instead of the outgoing ones.
@khwajawisal1220
@khwajawisal1220 4 жыл бұрын
its using einstein notation, not the normal one that we use.
@kristofneys2349
@kristofneys2349 4 жыл бұрын
There are many mistakes or confusing comments in this presentation, no wonder the audience keeps asking questions. Not a good talk at all....
@mehmetf.demirel8647
@mehmetf.demirel8647 4 жыл бұрын
great talk! the audience questions were helpful, but i felt like they were a bit too many in that they kinda negatively affected the flow of the talk.
@ashutoshukey3803
@ashutoshukey3803 3 жыл бұрын
The explanation ability and use of high-level diagrams by the presenter were phenomenal. Questions from the audience definitely messed up the flow of the presentation quite a bit though.
@MobileComputing
@MobileComputing 4 жыл бұрын
While the audience questions were mildly irritating (to put it, mildly), bombarding the speaker during his intro with questions that could reasonably be expected to be answered eventually in an 1-hour talk, why would the speaker give a talk on one of the most advanced neural network architecture to an audience without any machine learning background?
@ziranshuzhang6831
@ziranshuzhang6831 4 жыл бұрын
You are right, I am expecting to quickly adopt the GNN concept but the audience keeps asking irritating questions that I have to constantly hit the right button.
@ohmyumbrella
@ohmyumbrella 3 жыл бұрын
I agree. I mean I do see the point of giving this lecture to an audience without previous exposure to ML , if it is for the purpose of attracting them to the subject but in that case there should have been another video of the same lecture without so much interruption. It would take extra time and effort but for people who are trying to effectively learn GNN and have some knowledge of basic ML, these questions are very annoying and hinders the learning experience.
@robbat1209
@robbat1209 3 ай бұрын
The questions of the audience are absolutely valid, if they bother you there are plenty of other videos without an audience that you could watch
@MobileComputing
@MobileComputing 3 ай бұрын
@@robbat1209 This comment is absolutely valid. If it bothers you, there are plenty of other comments that you could read. This talk was from 4 years ago. This was one of the only sources about GNN, and before GenAI video summaries would allow audience like myself to comfortably skip ahead without fear of missing important information.
@robbat1209
@robbat1209 3 ай бұрын
@@MobileComputing 🤣🤣 true, my bad. The questions weren’t too horrible tho
@rembautimes8808
@rembautimes8808 3 жыл бұрын
"Spherical Cow" - funniest analogy yet for a Neural Net layers. Great talk
@alphaO27
@alphaO27 3 жыл бұрын
In 16:51, I think he meant for each node connecting to n (instead of n_j), because from the expression, we take all nodes n_j connected to n to be able to calculate the new state of node n h_t^n.
@runggp
@runggp 3 жыл бұрын
awesome talk! The MSR audience asked quite a few questions, which are actually helpful , eg, what are they, how they work/update, why they are created and designed this way, etc
@shawnz9833
@shawnz9833 3 жыл бұрын
can you help with that: what is a MLP in "you multiply it with a single layer MLP" @23:29?
@Peaceluvr18
@Peaceluvr18 3 жыл бұрын
Wow, what an excellent presentation, from someone with an ML background. Explains the basics a bit but also covers deep concepts. Super clear graphics! Seriously whoever made the graphics for this can I hire you to do my slide graphics? And thought it was very cool that the lecture attendees were bold enough to ask so many questions! Wish people asked more questions during my lectures+talks.
@heejuneAhn
@heejuneAhn 3 жыл бұрын
He explains using time progress, which make some cofusion to the audience and me.
@jonfe
@jonfe 8 ай бұрын
I dont understand why GRU is used, the input in GRU is a (Node x Caracteristics) Matrix, where is the temporal dimension?
@arjitarora8418
@arjitarora8418 8 ай бұрын
Great Introduction!
@instantpotenjoyer
@instantpotenjoyer 4 жыл бұрын
DiscussIon on the actual topic starts ~ 6:40
@pharofx5884
@pharofx5884 4 жыл бұрын
dat cough frequency suspiciously high. Distance thyself socially sir.
@maloxi1472
@maloxi1472 4 жыл бұрын
Don't worry, it's from November 2019
@jmplaza4947
@jmplaza4947 4 жыл бұрын
@@maloxi1472 COVID was already spreading then, right? I hope he's ok... wherever he is now...
@danielliu9616
@danielliu9616 4 жыл бұрын
He is maybe THE patient zero
@yanzexu6912
@yanzexu6912 4 жыл бұрын
bless him
@stackexchange7353
@stackexchange7353 3 жыл бұрын
@@danielliu9616oh shii
@minghan111
@minghan111 4 жыл бұрын
too many questions, just wait for the speaker please.
@pharofx5884
@pharofx5884 4 жыл бұрын
In formulating their questions they re-explained what is going on an order of magnitude better than the speaker. thats kinda sad
@MobileComputing
@MobileComputing 4 жыл бұрын
@@pharofx5884 I just watched the version from 2 years ago. Only 18 minutes long, but almost identical in content, yet that was much clearer. Really sad to see.
@eljangoolak
@eljangoolak 3 жыл бұрын
Very good presentation, but it is very difficult to follow with all the interrupting questions
@mansurZ01
@mansurZ01 4 жыл бұрын
35:40 I think the dimensionality of M should be (num_nodes x D), unless D==M. EDIT: from what follows, it should be M = HE, and D can be different from M.
@arisioz
@arisioz 2 жыл бұрын
Are those actual MS employers in the crowd? They are worse than 1st year CS students
@peter-holzer-dev
@peter-holzer-dev Жыл бұрын
I am pretty sure this is a great talk but unfortunately all the questions in between disturbs the flow a lot (also because most of them are hard to understand acoustically).
@sm_xiii
@sm_xiii 4 жыл бұрын
Where can we get the slide deck please?
@soniagupta2851
@soniagupta2851 2 жыл бұрын
Excellent explanation
@goblue1011
@goblue1011 2 жыл бұрын
Honestly some of the audience who raised questions have quite big ego and have no idea what they are talking about.
@r.dselectronics3349
@r.dselectronics3349 4 жыл бұрын
i am a researcher..the video contain the beautiful concept ...i like very much....:)..specially binary classification part ...i am so excited about this concepts....
@Exhora
@Exhora Жыл бұрын
29:35 about CGNs, he said you multiply the sum of the messages with your own state. But in the equation, it is a sum. I didn't get which one is correct.
@pielang5524
@pielang5524 3 жыл бұрын
Let the speaker talk!
@vritansh14
@vritansh14 3 ай бұрын
Amazin Talk !
@losiu998
@losiu998 3 жыл бұрын
Great presentation. But I have to point out something. I have no idea why you would use einstein notation instead of simple matrix multiplication? It raises unnecessary confusion and it's not related to GNNs.
@lidiias5976
@lidiias5976 3 ай бұрын
36:53 what is M in the shape (num_nodes by M)?
@DED_Search
@DED_Search 3 жыл бұрын
is there anyway to get access to the slides? Great talk! Thanks.
@aladdin1776
@aladdin1776 3 ай бұрын
The talk is quite intriguing; however, the interruptions from the audience are quite disturbing.
@FsimulatorX
@FsimulatorX 3 жыл бұрын
OMG! For a second I thought he looked like the CEO of Google and was wondering to myself: why would the CEO of Google do a presentation about Neural Networks AT MICROSOFT!!
@sebamurgui
@sebamurgui 4 жыл бұрын
OH MY GOD that audience
@miketurner3461
@miketurner3461 4 жыл бұрын
I had to stop watching because of them
@kellybrower301
@kellybrower301 3 жыл бұрын
I wanted to believe it wasn’t true.
@sunaxes
@sunaxes 2 жыл бұрын
So GNN is just message passing on a graph or did I miss something? This has been around since way back, isnt it??
@BapiKAR
@BapiKAR 4 жыл бұрын
could you please post the ppt here? Thanks
@mohammedamraoui4147
@mohammedamraoui4147 7 ай бұрын
and for Edge classification?
@codewithyouml8994
@codewithyouml8994 3 жыл бұрын
The miss, at 40:00 was right .... as i was alsoooo really confused, like all the matrix operations were seemed to be invalid if not swapped ... lol what kind of inverted conventions are these ....
@samuisuman
@samuisuman 4 жыл бұрын
There is no intuitive explaination, but quite informative
@klaimouad740
@klaimouad740 3 жыл бұрын
can we propagate messages for example depending on the edges features for example if the distance from node n to m is greater than their distance to p then we propagate the message first to p then we perform the propagation to the other node m
@NewtonInDaHouseYo
@NewtonInDaHouseYo 2 жыл бұрын
Excellent introduction, thanks a lot!
@hamishhall5423
@hamishhall5423 4 жыл бұрын
What is the dimension M, for the msg_to_be_sent and received_messages etc. I get that D is the dimension of the node representation, N the num_nodes etc
@BruceChen27
@BruceChen27 3 жыл бұрын
Seems several people were not healthy
@leventesipos1456
@leventesipos1456 Ай бұрын
The notation is incomplete or incorrect in so many places on the presentation, that it was hard to follow.
@hfkssadfrew
@hfkssadfrew 3 жыл бұрын
34:46 . it is NOT A* N, it is N' * A ....
@giannismanousaridis4010
@giannismanousaridis4010 4 жыл бұрын
I found the slide for everyone who asked here: miltos.allamanis.com/files/slides/2020gnn.pdf (Idk If I'm not supposed or allowed to post the link here, if not sorry for that, I'll delete my comment. Just let me know).
@warpdrive9229
@warpdrive9229 4 ай бұрын
You can let it remain here :)
@patrickadjei9676
@patrickadjei9676 3 жыл бұрын
People are not happy for the many questions. However, I'm kinda sad that he doesn't re-state the questions before answering :( like why?
@halilibrahimakgun7569
@halilibrahimakgun7569 Жыл бұрын
Can you share the slides please. I like them.
@arashjavanmard5911
@arashjavanmard5911 3 жыл бұрын
It would be great if you could also publish the slides!
@mayankgolhar8761
@mayankgolhar8761 3 жыл бұрын
Slides from the presenter's website: miltos.allamanis.com/files/slides/2020gnn.pdf
@arashjavanmard5911
@arashjavanmard5911 3 жыл бұрын
@@mayankgolhar8761 thanks
@yb801
@yb801 3 жыл бұрын
Is this a 2016 talk?
@pielang5524
@pielang5524 3 жыл бұрын
at time 35:51, the adjacency matrix A should also depend on edge type k imo.
@pielang5524
@pielang5524 3 жыл бұрын
OK.. The presenter confirmed this shortly after...
@sherazbaloch1642
@sherazbaloch1642 2 жыл бұрын
Need more tutorial on GNN
@khwajawisal1220
@khwajawisal1220 4 жыл бұрын
thats why software engineers should not teach, you assume everything is a design/modeling detail when in reality they are part of the mathematics behind them. and i seriously miss those old days where professors used to teach with a chalk and board.
@blanamaxima
@blanamaxima 3 жыл бұрын
Is this related to bread baking ?
4 жыл бұрын
Where is the inputs and outputs?
@miketurner3461
@miketurner3461 4 жыл бұрын
Clearly you would've been one of the people asking foolish questions they could answer using Google
@RARa12812
@RARa12812 3 жыл бұрын
How tp turn off questions
@ZephyrineFreiberg
@ZephyrineFreiberg 3 жыл бұрын
It should be (A^T)*N
@2000sunnybunny
@2000sunnybunny 4 жыл бұрын
Great session !
@rocking4joy
@rocking4joy 3 жыл бұрын
Don't understand the praise in the comment section, I actually found it kind of sloppy with typo-s but the audience and the questions are really great.
@lifelemons8176
@lifelemons8176 3 жыл бұрын
horror crowd. this is something I see in every microsoft talk
@WeeeAffandi
@WeeeAffandi Жыл бұрын
You can tell this lecture was recorded during the prime Covid by hearing the constant coughing from audience (and the speaker)
@carloderamo
@carloderamo Жыл бұрын
Some good person should take this video and remove all the awful questions from the audience
@CodingwithIndy
@CodingwithIndy 4 жыл бұрын
Is that SPJ I hear in the audience at 9:18
@losiu998
@losiu998 3 жыл бұрын
34:22 - is it vector-matrix multiplication? if so, the result is wrong i guess @edit: Matrix A should have ones under diagonal, not above - then result is as presented
@aichemozenerator8446
@aichemozenerator8446 2 жыл бұрын
good
@ngideo
@ngideo 4 жыл бұрын
I hear Simon Peyton Jones in the audience
@rherrmann
@rherrmann 3 жыл бұрын
Easily recognizable indeed!
@kutilkol
@kutilkol 3 жыл бұрын
Rip ears. Wtf with the caughing. Use at least some compressor for the vocal audio omg.
@JK-sy4ym
@JK-sy4ym 4 ай бұрын
Unfortunately many good researchers can’t present their work well.
@guest1754
@guest1754 Жыл бұрын
Wondering how many people had covid in that recording...
@sirisaksirisak6981
@sirisaksirisak6981 3 жыл бұрын
At least it save time in doing strategy.
@dalahmah
@dalahmah 4 ай бұрын
I found the lecture’s atmosphere dull and depressing. It seems that the lecturer was forced to give the lecture!
@barriesosinsky9566
@barriesosinsky9566 3 жыл бұрын
An aromatic ring is not a "single bone" next to a "double bone." The bonds are a resonance form in a single state. Treating them with graph theory is not supported by current models.
@yanyipu4029
@yanyipu4029 3 жыл бұрын
Great Video but annoying audience
@aglaiawong8058
@aglaiawong8058 3 жыл бұрын
the interruptions are so annoying...
@KeshavDial
@KeshavDial 2 жыл бұрын
The audience ruined this presentation. I have never felt worse for a presenter.
@kushalneo
@kushalneo 3 жыл бұрын
👍
@jebinjames6317
@jebinjames6317 3 жыл бұрын
People need to remember they’re watching a free video on KZbin...it’s not your advanced ML private tutoring session...
@ММ_4321
@ММ_4321 3 жыл бұрын
Kind of confusing for me. And the audience very annoying
@thevanisher4609
@thevanisher4609 9 ай бұрын
Horrible audience, great talk!
@iSpades0
@iSpades0 3 жыл бұрын
that audience was pretty annoying tbh
@yiweijiang
@yiweijiang 3 жыл бұрын
So that's machine learning! Haha, lol
@eaglesofmai
@eaglesofmai 3 жыл бұрын
are GNN's patented? does anyone know if using a paritcular ANN construct can be subject to litigation?
@miketurner3461
@miketurner3461 4 жыл бұрын
The audience needs to take a freaking ML 101 class before asking stupid questions
@williamashbee
@williamashbee 3 жыл бұрын
cough totally ruined the presentation.
@MrArmas555
@MrArmas555 4 жыл бұрын
++
@Tyomas1
@Tyomas1 3 жыл бұрын
awful introduction
@robodoc1446
@robodoc1446 3 жыл бұрын
Appalling talk! It shows why coders are terrible in public speaking or often fail to explain things in a transparent manner. Before explaining how message passing is done in an end-to-end learning architecture, he jumped to talk about Gated GNN leaving an impression that GRU may be an important part of GNN. This is one of the reasons why he got so many questions and confusion surrounding his lecture.....what is h_t? "well, this is not something that changes"... seriously Microsoft!
@TheDavidlloydjones
@TheDavidlloydjones 2 жыл бұрын
Word salad. A hopeless mess of talking at and around a topic without actually touching it. Take it down. Tell the guy to try again.
@saulrojas2679
@saulrojas2679 2 жыл бұрын
Slides can be found at: miltos.allamanis.com/files/slides/2020gnn.pdf
GEOMETRIC DEEP LEARNING BLUEPRINT
3:33:23
Machine Learning Street Talk
Рет қаралды 204 М.
Intro to graph neural networks (ML Tech Talks)
51:06
TensorFlow
Рет қаралды 179 М.
Random Emoji Beatbox Challenge #beatbox #tiktok
00:47
BeatboxJCOP
Рет қаралды 32 МЛН
Cool Parenting Gadget Against Mosquitos! 🦟👶 #gen
00:21
TheSoul Music Family
Рет қаралды 33 МЛН
Человек паук уже не тот
00:32
Miracle
Рет қаралды 2,9 МЛН
Theoretical Foundations of Graph Neural Networks
1:12:20
Petar Veličković
Рет қаралды 92 М.
Deep Learning Interview Prep Course
3:59:50
freeCodeCamp.org
Рет қаралды 424 М.
Graph Neural Networks - a perspective from the ground up
14:28
MIT Introduction to Deep Learning | 6.S191
1:09:58
Alexander Amini
Рет қаралды 681 М.
Something Strange Happens When You Take This To Its Logical Conclusion
32:44
Liquid AI: Product Launch Webcast 10/23
3:06:11
Liquid AI
Рет қаралды 1,5 М.
How Deep Neural Networks Work - Full Course for Beginners
3:50:57
freeCodeCamp.org
Рет қаралды 4,2 МЛН
Graph Neural Networks: A gentle introduction
29:15
Aladdin Persson
Рет қаралды 45 М.
Graph Representation Learning (Stanford university)
1:16:53
Machine Learning TV
Рет қаралды 95 М.
Random Emoji Beatbox Challenge #beatbox #tiktok
00:47
BeatboxJCOP
Рет қаралды 32 МЛН