prepare.args: Generic function for preparing the sgl call arguments

Description Usage Arguments Value Author(s) See Also

View source: R/prepare_args.R

Description

Compute and prepare the sgl call arguments for the objective function

\mathrm{loss}(\mathrm{data})(β) + λ ≤ft( (1-α) ∑_{J=1}^m γ_J \|β^{(J)}\|_2 + α ∑_{i=1}^{n} ξ_i |β_i| \right)

where \mathrm{loss} is a loss/objective function. The n parameters are organized in the parameter matrix β with dimension q\times p. The vector β^{(J)} denotes the J parameter group, the dimension of β^{(J)} is denote by d_J. The dimensions d_J must be multiple of q, and β = (β^{(1)} \cdots β^{(m)}). The group weights γ \in [0,∞)^m and the parameter weights ξ \in [0,∞)^{qp}.

Usage

1

Arguments

data

a data object

...

additional parameters

Value

block_dim

a vector of length m, containing the dimensions d_J of the groups (i.e. the number of parameters in the groups)

groupWeights

a vector of length m, containing the group weights

parameterWeights

a matrix of dimension q \times p, containing the parameter weights

alpha

the α value

data

the data parsed to the loss module

group_order

original order of the columns of β. Before sgl routines return β will be reorganized according to this order.

Author(s)

Martin Vincent

See Also

prepare.args.sgldata

Other sgldata: add_data.sgldata, create.sgldata, prepare.args.sgldata, prepare_data, rearrange.sgldata, subsample.sgldata


sglOptim documentation built on Oct. 21, 2018, 9:04 a.m.