VahidooX
The loss function and the implementation of 1st&2rd order
I don't understand the implementation very well. It seems that the loss function is quite different from what given in the paper. Also, the paper gives two methods, 1st order and 2rd order, is this implementation just include the first order method. I'm an undergraduate student new in this field. Hope to get some instructions.
Hi,
Sorry for the late reply. I should have missed your question. I am not sure if you still have this question.
Consider that in the paper they have simplified their optimization problems in section 4.2. My code is similar to order 2 where they have two embeddings for each vertex (one for the node itself, one for the context). For order 1, they have just one set of embeddings. Anyway, they are very close to each other. Moreover, their implementation in C has some differences with the paper. For example they did not use the log in the loss. But all these implementations are very close to each other and should work.
In the next two months, I will add both orders and update this code to be consistent with their original implementation exactly.
VahidooX, have you updated the code as you mentioned in your comment? Also, what are the pre-requisites for running your code?
