Titouan Vayer
Titouan Vayer
Hello Fabian, Thank you for your answer, yes will do that ! Titouan Le jeu. 12 mai 2022 à 21:40, Fabian Pedregosa ***@***.***> a écrit : > yeah, I think...
Hello, I think it is because your "comp_(v)" is a function and in our code it should be a string (the name of the metric as written in the documentation...
Hi, Yes indeed that was a poor choice at that time to use a custom graph dataset.. I think this should be easily fixable. The only problem I see is...
Alright, maybe I am wrong but I suppose it is because you do not have node attributes in your graphs. PSCN works with attributes. For some datasets where the nodes...
Hi thank you for this comment. Indeed I believe this is because the GW (and FGW) distance is a non-convex problem, so the optimisation may fall into some bad local...
I think you have to recompile properly the pynauty library, since the version here is compiled for Mac and Python3.5. I am not a linux expert but maybe the python...
Terrific thanks !
Dear Bion, Indeed this is a problem but the main reason is that GW1D is not defined for distributions with not the same number of atoms yet. The reason is...
Yes exactly it is beacause the assignement problem. And yes this paper introduce some works to circumvent this limitation for the sliced wasserstein distance! however it remains to be proved...
Yes it is possible and would be interesting to see the results !