Machinelearning: StochasticDualCoordinateAscentClassifier
anyone know how is "StochasticDualCoordinateAscentClassifier()" working? is there any explanation for that? I cant understand how it's working in the background.
Is it a neural net model or some other model?
bs6523
All 4 comments
I take it from this paper that it looks like it's a way to perform a Support Vector Machines (SVM) algorithm.
jwood803
on 19 Jun 2018
Hi @jwood803 and @bs6523 .
The method certainly can be applied as an SVM if you apply the method on the binary classifier with the HingeLossSDCAClassificationLossFunction. However by default it is logistic loss, that is, LogLossSDCAClassificationLossFunction.
There's a paper here about this method in particular.
...aaaand despite the fact that my name is right there in the paper and all they had to do was copy-paste it, I see whoever put this page together still managed to misspell my name "Finely." That's a great start to my day. :smile:
TomFinley
on 19 Jun 2018
@bs6523,
Simply put, SDCA is a fast, scalable, multicore linear classifier with support for L1 and L2 regularization.
glebuk
on 19 Jun 2018
DRI RESPONSE: Questions seems to have been answered. I plan to close the issue withing the next few days.
abgoswam
on 18 Oct 2018
Related issues
rogancarr
路
4Comments
daholste
路
4Comments
dev8546
路
3Comments
rebecca-burwei
路
3Comments
OneCyrus
路
4Comments
Most helpful comment
Hi @jwood803 and @bs6523 .
The method certainly can be applied as an SVM if you apply the method on the binary classifier with the
HingeLossSDCAClassificationLossFunction. However by default it is logistic loss, that is,LogLossSDCAClassificationLossFunction.There's a paper here about this method in particular.
...aaaand despite the fact that my name is right there in the paper and all they had to do was copy-paste it, I see whoever put this page together still managed to misspell my name "Finely." That's a great start to my day. :smile: