Imagine that I want to regularize the same column more than once, with different regularizers. For example, if I wanted to perform Non negative matrix factorization with quadratic regularization I would have to use NonNegConstraint and QuadReg for every column of X and every column of Y. Right now that is not possible. Would it be very hard to implement it?
Imagine that I want to regularize the same column more than once, with different regularizers. For example, if I wanted to perform Non negative matrix factorization with quadratic regularization I would have to use
NonNegConstraintandQuadRegfor every column ofXand every column ofY. Right now that is not possible. Would it be very hard to implement it?