[SAM] What is the guidance to choose a proper graph_sampler?

FenTechSolutions / CausalDiscoveryToolbox

Package for causal inference in graphs and in the pairwise settings. Tools for graph structure recovery and dependencies are included.

https://fentechsolutions.github.io/CausalDiscoveryToolbox/html/index.html

MIT License

1.08k stars 198 forks source link

[SAM] What is the guidance to choose a proper graph_sampler? #101

Closed ConnLiu closed 3 years ago

ConnLiu commented 3 years ago

@Diviyan-Kalainathan
As you codes show, there are different graph_sampler like SimpleMatrixConnection\ MatrixSampler(Gumble)\MatrixSampler2\MatrixSampler3. So I am wondering how to choose a proper one? And why you create various graph_sampler?

diviyank commented 3 years ago

Hello @ConnLiu,

These different graph samplers originate from how they are sampled (gumble softmax, etc...) ; usually, it is recommended to leave this parameter as default ; this parameter was used in development and cross-evaluated. This is kept for retro-compatibility, but we might retrain this parameter in the near future.

Hope this answers your question; Best, Diviyan

ConnLiu commented 3 years ago

Got it~ Thanks a lot! By the way, are you planning to submit the SAM paper to a conference or journal? I think it would be nice to see this work published!

diviyank commented 3 years ago

Thanks for the kind words :) The paper is in the process of submission for JMLR!

Best, Diviyan