RegReg — RegReg Docuentation - GitHub Pages
RegReg is a proximal gradient algorithm Python framework for prototyping and solving regularized regression problems such as the LASSO. The goal is to enable practitioners to quickly and easily experiment with a variety of different models and choices of regularization.
RegReg — RegReg Docuentation - GitHub Pages
RegReg is designed to solve the generic problem \[\mbox{minimize}_\beta \quad \mathcal{L}(\beta) + \mathcal{P}(\beta)\] where \(\mathcal{P}\) is a convex function with a simple proximal map.
RegReg — RegReg Docuentation - GitHub Pages
API¶ Release. v0.1.3+9.g60e3aef. Date. September 24, 2019. affine. Module: affine Classes. AffineError; adjoint; affine_sum; affine_transform; aslinear
RegReg — RegReg Docuentation - GitHub Pages
RegReg documentation¶ Release. v0.1.3+9.g60e3aef. Date. September 24, 2019. Download PDF. Contents:
RegReg — RegReg Docuentation
simple_problem ¶ class regreg.problems.simple.simple_problem (smooth_atom, proximal_atom) ¶. Bases: regreg.problems.composite.composite __init__ (smooth_atom ...
RegReg — RegReg Docuentation - GitHub Pages
class regreg.smooth.affine_smooth (smooth_atom, atransform, store_grad=True, diag=False) ¶ Bases: regreg.smooth.smooth_atom. Composition of a smooth objective with an affine transform. __init__ (smooth_atom, atransform, store_grad=True, diag=False) ¶ Parameters. smooth_atom: regreg.smooth.smooth_atom
RegReg — RegReg Docuentation - GitHub Pages
The basic atoms in regreg know what their conjugate is. Our hinge loss, hinge_rep , is the composition of an atom , and an affine transform. This affine transform is split into two pieces, the linear part, stored as linear_transform and its offset stored as atom.offset .
RegReg — RegReg Docuentation - GitHub Pages
Downloading and installing RegReg¶ The RegReg source code is hosted at. http://github.com/regreg/regreg. RegReg depends on the following Python tools: numpy. scipy. Cython. You can clone the RegReg github repo using:
RegReg — RegReg Docuentation
regreg.affine.fused_lasso.difference_transform (X, order=1, sorted=False, transform=False) ¶ Compute the divided difference matrix for X after sorting X. Parameters. X: np.array, np.float, ndim=1
RegReg — RegReg Docuentation
If True, Nesterov weights are restarted every time the objective value increases