scikit-network

Graph software

[Johannes Gasteiger, né Klicpera @gasteigerjo]

Why is anyone comparing runtime to NetworkX graph algorithms? NetworkX uses pure Python implementations, _everyone_ can be orders of magnitude faster than that!

Try beating graph-tool, scikit-network or cugraph instead. (Good luck with that!)

[scikit-network @ScikitN]

Scikit-network now comes with a #publication attached!
Find our article in the last issue of the @JmlrOrg :
https://t.co/o3m5CHayO8
Thanks to all the contributors and to the reviewers!

[scikit-network @ScikitN]

20 years ago, the awesome paper "Graph structure in the Web" was published https://t.co/ITLu4vEEnP
Did things change in between ? Is it still relevant today ?
@jure @misovalko @vagelispapalex @trekkinglemon @LIP6_lab @LINCS_Paris @lgetoor @thomaskipf @tinaeliassi @ladamic

[scikit-network @ScikitN]

Les #députés face à leurs #votes : Ce que nous apprend la théorie des #graphes.
@AssembleeNat @LaREM_AN @Republicains_An @FranceInsoumise @socialistesAN @EELV_Paris @DeputesUDI_Ind @MoDem @RNational_off @deputesPCF @GroupeLibTerrAN
https://t.co/sJ7GJ8k02r

[Gael Varoquaux 🦋 @GaelVaroquaux]

@behrenstimb Was thinking: we should have tracks with high requirements on quality of evidence, but less novelty (series B 😊).

For instance, we have a manuscript showing that HCQ has little effect on Covid19, but it is being rejected by medical journals, because it is considered known.

[scikit-network @ScikitN]

@emaros96 True. But is it the authors fault or does it simply reflect the lack of major open datasets in this field ?

[scikit-network @ScikitN]

Very proud of the #memory footprint of #scikitnetwork!

RAM usage for loading the #Orkut graph of #Konect (3M nodes, 117M edges) on a computer (32GB RAM):

#iGraph -> 18GB
#graphtool -> 10GB
#NetworkX -> memory overflow
#scikitnetwork -> 1GB

The #CSR format of #SciPy is awesome!

[scikit-network @ScikitN]

@apachaves @LINCS_Paris @alonsosilva @scikit_learn @TelecomParis_ @IP__Paris @HubFranceIA @ActuIAFr @Etalab @AI4EU @nicogla Thanks!

[scikit-network @ScikitN]

@dhimmel @ThosVarley @LINCS_Paris @jkbren @scikit_learn @TelecomParis_ @IP__Paris @HubFranceIA @ActuIAFr @Etalab @AI4EU @nicogla Correct. Thanks for the reply!

[Nicolas Glady @nicogla]

.@TelecomParis_ is very proud to announce a new version of scikit-network! A #python #opensource package for #ML with graphs using the same API as @scikit_learn. By Thomas Bonald, Bertrand Charpentier, Nathan de Lara et Quentin Lutz. https://t.co/ILOdeSuKww