3 ms·

Nature Communications paper link: https://www.nature.com/articles/s41467-024-53373-4 https://www.nature.com/articles/s41467-024-53373-4 GitHub link: https://gi

Nature Communications paper link: https://www.nature.com/articles/s41467-024-53373-4 https://www.nature.com/articles/s41467-024-53373-4 GitHub link: https://github.com/Computational-Turbulence-Group/SURD https://github.com/Computational-Turbulence-Group/SURD

Arech 2y ago
Thanks. This should be the actual submission instead of the marketing-speak blah-blah
- passwordoops 2y ago
  I'm in your camp, but I've found with my own submissions here, marketing-speak gets more engagement than the papers. So my compromise is to post the PR, but give the paper link in the first comment
BenoitP 2y ago
> Decomposition of causality: It decomposes causal interactions into redundant, unique, and synergistic contributions. Seen elsewhere: https://github.com/BCG-X-Official/facet https://github.com/BCG-X-Official/facet, which uses SHAP attributions as inputs: > The SHAP implementation is used to estimate the shapley vectors which FACET then decomposes into synergy, redundancy, and independence vectors. But FACET it still about sorting things out in the 'correlation world'. To get back to SURD: IMHO, when talking about causality one should incorporate some kind of precedence, or order; One thing is the cause of another. Here in SURD they sort of introduce it in a roundabout way by using time's order: > requiring only pairs of past and future events for analysis But maybe we could have had fully-fledged custom DAGs, like from here https://github.com/nathanwang000/Shapley-Flow https://github.com/nathanwang000/Shapley-Flow (which don't yet have the redundant/unique/synergistic decomposition) Also, how do we deal with undetectable "post hoc ergo propter hoc" fallacy, though? (travesting time as causal ordering). How do we deal with confounding? Custom DAGs would have been great. I'm longing for a SURD/SHAP/FACET/Shapleyflow integration paper. We're so close to it.

Arech 2y ago

Thanks. This should be the actual submission instead of the marketing-speak blah-blah

passwordoops 2y ago
I'm in your camp, but I've found with my own submissions here, marketing-speak gets more engagement than the papers. So my compromise is to post the PR, but give the paper link in the first comment

BenoitP 2y ago

> Decomposition of causality: It decomposes causal interactions into redundant, unique, and synergistic contributions. Seen elsewhere: https://github.com/BCG-X-Official/facet https://github.com/BCG-X-Official/facet, which uses SHAP attributions as inputs: > The SHAP implementation is used to estimate the shapley vectors which FACET then decomposes into synergy, redundancy, and independence vectors. But FACET it still about sorting things out in the 'correlation world'. To get back to SURD: IMHO, when talking about causality one should incorporate some kind of precedence, or order; One thing is the cause of another. Here in SURD they sort of introduce it in a roundabout way by using time's order: > requiring only pairs of past and future events for analysis But maybe we could have had fully-fledged custom DAGs, like from here https://github.com/nathanwang000/Shapley-Flow https://github.com/nathanwang000/Shapley-Flow (which don't yet have the redundant/unique/synergistic decomposition) Also, how do we deal with undetectable "post hoc ergo propter hoc" fallacy, though? (travesting time as causal ordering). How do we deal with confounding? Custom DAGs would have been great. I'm longing for a SURD/SHAP/FACET/Shapleyflow integration paper. We're so close to it.