Stanford CS224W: Machine Learning with Graphs | 2021 | Lecture 1.3 - Choice of Graph Representation​

[@hoholord5699]

amazing series to learn the connection between machine learning and graphs, thank you

[@andreasbeschorner1215]

Greetings, may I ask why you exclude loops from the adjacency matrix? Edit: I should have been more patient, loops come into play just a few minutes later...

[@wege8409]

Graph nns are definitely the way forward I feel

[@carolinemoraes3704]

Great class!! Thank you :)

[@lipeemacao1348]

Amazing class

[@gamerboygs153]

Good class

[@aneesharma3240]

@15:58 isnt in the matrix for the self loops, row 4 and column 4 value should be 2, as mentioned earlier the degree of a self loop is 2 and not 1. As a result the degree of node 4 will be 3 (which you are saying as 3 to be degree of node 4) but looking and summing at row 4 and column 4, the value it is coming out to be 2???????

[@prasanthkarunakaran5883]

Actually The adjacency matrix denotes that rows to the Nodes and columns to nodes that are connected to it. Let take row 1 as Node 1, then nodes which are connected to nodes 1 are Node 1 (self loop) and node 2 and node 3. Then the row will be (1, 1, 1, 0). The value 1 denotes only the connectivity status, not the count. I hope you got this!

[@TIENTI0000]

good course

[@sarbajit.g]

@1:48 what will be the effect of losing use-case-specific information?