You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
As recently demonstrated by Montagna et al. (2023), score matching algorithms like SCORE, DAS and NoGAM might overperform other algorithms under certain assumption violations.
This makes them attractive candidates for practical applications, especially under some relatively mild assumptions regarding the data generating process.
I believe they would be a great extension of gCastle's unique algorithm arsenal.
Curious to hear your thoughts.
The text was updated successfully, but these errors were encountered:
After taking a quick look, these algorithms seem like they could be worthwhile to include. There are what appear to be complete implementations of the algorithms available at https://github.com/py-why/dodiscover under an MIT license which should be possible to adapt to Castle. We will look further into it but I can't give any timeline as we are all quite busy these days. If you want to take a shot at integrating the algorithms you are more than welcome :)
Hi team,
As recently demonstrated by Montagna et al. (2023), score matching algorithms like SCORE, DAS and NoGAM might overperform other algorithms under certain assumption violations.
This makes them attractive candidates for practical applications, especially under some relatively mild assumptions regarding the data generating process.
I believe they would be a great extension of gCastle's unique algorithm arsenal.
Curious to hear your thoughts.
The text was updated successfully, but these errors were encountered: