Open homerours opened 3 years ago
That would be a very nice addition. However, while non-negativity constraints are very straightforward for LASSO, it is not straightforward to implement non negative group lasso.
It seems like we at least would need a different optimisation algorithm than FISTA, such as ADMM. Unfortunately, that would be a substantial effort to implement and test, so I don't think I have the time to implement that.
Hi, thank you for this nice repo :) Is there any way to use it to do positive group lasso, that is minimizing over vector with non-negative coefficients ? In scikit-learn, there is a
positive=True
parameter that handles this. Is there such possibility here, or is there an easy way to solve this ? Thank you!