skprox._pgd
Module Contents
Classes
This class will be used by all proximal gradient descent algorithms. |
Attributes
- skprox._pgd.PROXIMAL_OPERATORS
- skprox._pgd.PROXIMAL_PARAMS
- class skprox._pgd._PGDMixin
This class will be used by all proximal gradient descent algorithms.
We would like it to flexibly allow for different gradient descent algorithms including gradient descent with momentum, Nesterov’s accelerated gradient descent, and stochastic gradient descent.
- _proximal_gradient_descent(X, y, coef_)
This is the main proximal gradient descent algorithm.
- abstract _compute_gradient(X, y, coef_)
- abstract _objective(X, y, coef_)