function [X, C]=pathSolutionLogistic(A, y, z, opts)
%
%% Fuction pathSolution:
% Solving the pathwise solutions
%
%% Input & Output parameters
% See the description of the related functions
%
%% Copyright (C) 2009-2010 Jun Liu, and Jieping Ye
%
% You are suggested to first read the Manual.
%
% For any problem, please contact with Jun Liu via j.liu@asu.edu
%
% Last modified 2 August 2009.
%
% Related functions:
% sll_opts,
% eppVector, eppMatrix, eplb,
%
% LogisticR, LogisticC,
% nnLogisticR, nnLogisticC
% glLogisticR, mtLogisticR, mclLogisticR
%%
switch(opts.fName)
case 'LogisticR'
z_num=length(z); % the number of parameters
[z_value, z_ind]=sort(-z); % sort z in a decresing order
z_value=-z_value; % z_value in a decreasing order
n=size(A,2); % the dimensionality of the data
X=zeros(n,z_num); % set the size of output X
C=zeros(1,z_num); % set the size of output C
% run the code to compute the first solution
[x, c, funVal]=LogisticR(A, y, z_value(1), opts);
X(:,z_ind(1))=x; % store the solution x
C(1,z_ind(1))=c; % store the solution c
% set .init for warm start
opts.init=1; % using the defined ones
for i=2:z_num
opts.x0=x; % warm-start of x
opts.c0=c; % warm-start of c
% run the function LogisticR
[x, c, funVal]=LogisticR(A, y, z_value(i), opts);
X(:,z_ind(i))=x; % store the solution x
C(1,z_ind(i))=c; % store the solution c
end
case 'LogisticC'
z_num=length(z); % the number of parameters
[z_value, z_ind]=sort(z); % sort z in an ascending order
n=size(A,2); % the dimensionality of the data
X=zeros(n,z_num); % set the size of output X
C=zeros(1,z_num); % set the size of output C
% run the code to compute the first solution
[x, c, funVal]=LogisticC(A, y, z_value(1), opts);
X(:,z_ind(1))=x; % store the solution x
C(1,z_ind(1))=c; % store the solution c
% set .init for warm start
opts.init=1; % using the defined ones
for i=2:z_num
opts.x0=x; % warm-start of x
opts.c0=c; % warm-start of c
% run the function LogisticC
[x, c, funVal]=LogisticC(A, y, z_value(i), opts);
X(:,z_ind(i))=x; % store the solution x
C(1,z_ind(i))=c; % store the solution c
end
case 'glLogisticR'
z_num=length(z); % the number of parameters
[z_value, z_ind]=sort(-z); % sort z in a decresing order
z_value=-z_value; % z_value in a decreasing order
n=size(A,2); % the dimensionality of the data
X=zeros(n,z_num); % set the size of output X
C=zeros(1,z_num); % set the size of output C
% run the code to compute the first solution
[x, c, funVal]=glLogisticR(A, y, z_value(1), opts);
X(:,z_ind(1))=x; % store the solution x
C(1, z_ind(1))=c; % store the solution c
% set .init for warm start
opts.init=1; % using the defined ones
for i=2:z_num
opts.x0=x; % warm-start of x
opts.c0=c; % warm-start of c
% run the function glLogisticR
[x, c, funVal]=glLogisticR(A, y, z_value(i), opts);
X(:,z_ind(i))=x; % store the solution x
C(1, z_ind(i))=c; % store the solution c
end
case 'mtLogisticR'
z_num=length(z); % the number of parameters
[z_value, z_ind]=sort(-z); % sort z in a decresing order
z_value=-z_value; % z_value in a decreasing order
n=size(A,2); % the dimensionality of the data
k=length(opts.ind)-1; % the number of tasks
X=zeros(n,k,z_num); % set the size of output X
C=zeros(z_num, k); % set the size of output C
% run the code to compute the first solution
[x, c, funVal]=mtLogisticR(A, y, z_value(1), opts);
X(:,:,z_ind(1))=x; % store the solution x
C(z_ind(1), :)=c; % store the solution c
% set .init for warm start
opts.init=1; % using the defined ones
for i=2:z_num
opts.x0=x; % warm-start of x
opts.c0=c; % warm-start of c
% run the function mtLogisticR
[x, c, funVal]=mtLogisticR(A, y, z_value(i), opts);
X(:,:, z_ind(i))=x; % store the solution x
C(z_ind(i), :)=c; % store the solution c
end
case 'mcLogisticR'
z_num=length(z); % the number of parameters
[z_value, z_ind]=sort(-z); % sort z in a decresing order
z_value=-z_value; % z_value in a decreasing order
n=size(A,2); % the dimensionality of the data
k=size(y,2); % the number of classes (tasks)
X=zeros(n,k,z_num); % set the size of output X
C=zeros(z_num, k); % set the size of output C
% run the code to compute the first solution
[x, c, funVal]=mcLogisticR(A, y, z_value(1), opts);
X(:,:,z_ind(1))=x; % store the solution x
C(z_ind(1), :)=c; % store the solution c
% set .init for warm start
opts.init=1; % using the defined ones
for i=2:z_num
opts.x0=x; % warm-start of x
opts.c0=c; % warm-start of c
% run the function mcLogisticR
[x, c, funVal]=mcLogisticR(A, y, z_value(i), opts);
X(:,:, z_ind(i))=x; % store the solution x
C(z_ind(i), :)=c; % store the solution c
end
case 'nnLogisticR'
z_num=length(z); % the number of parameters
[z_value, z_ind]=sort(-z); % sort z in a decresing order
z_value=-z_value; % z_value in a decreasing order
n=size(A,2); % the dimensionality of the data
X=zeros(n,z_num); % set the size of output X
C=zeros(1,z_num); % set the size of output C
% run the code to compute the first solution
[x, c, funVal]=nnLogisticR(A, y, z_value(1), opts);
X(:,z_ind(1))=x; % store the solution x
C(1,z_ind(1))=c; % store the solution c
% set .init for warm start
opts.init=1; % using the defined ones
for i=2:z_num
opts.x0=x; % warm-start of x
opts.c0=c; % warm-start of c
% run the function LogisticR
[x, c, funVal]=nnLogisticR(A, y, z_value(i), opts);
X(:,z_ind(i))=x; % store the solution x
C(1,z_ind(i))=c; % store the solution c
end
case 'nnLogisticC'
z_num=length(z); % the number of parameters
[z_value, z_ind]=sort(z); % sort z in an ascending order
n=size(A,2); % the dimensionality of the data
X=zeros(n,z_num); % set the size of output X
C=zeros(1,z_num); % set the size of output C
% run the code to compute the first solution
[x, c, funVal]=nnLogisticC(A, y, z_value(1), opts);
X(:,z_ind(1))=x; % store the solution x
C(1,z_ind(1))=c; % store the solution c
% set .init for warm start
opts.init=1; % using the defined ones
for i=2:z_num
opts.x0=x; % warm-start of x
opts.c0=c; % warm-start of c
% run the function LogisticC
[x, c, funVal]=nnLogisticC(A, y, z_value(i), opts);
X(:,z_ind(i))=x; % store the solution x
C(1,z_ind(i))=c; % store the solution c
end
case 'mtLogisticC'
z_num=length(z); % the number of parameters
[z_value, z_ind]=sort(z); % sort z in an ascending order
n=size(A,2); % the dimensionality of the data
k=length(opts.ind)-1; % the number of tasks
X=zeros(n,k,z_num); % set the size of output X
C=zeros(z_num, k); % set the size of output C
% run the code to compute the first solution
[x, c, funVal]=mtLogisticC(A, y, z_value(1), opts);
X(:,:,z_ind(1))=x; % store the solution x
C(z_ind(1), :)=c; % store the solution c
% set .init for warm start
opts.init=1; % using the defined ones
for i=2:z_num
opts.x0=x; % warm-start of x
opts.c0=c; % warm-start of c
% run the function mtLogisticC
[x, c, funVal]=mtLogisticC(A, y, z_value(i), opts);
X(:,:, z_ind(i))=x; % store the solution x
C(z_ind(i), :)=c; % store the solution c
end
case 'mcLogisticC'
z_num=length(z); % the number of parameters
[z_value, z_ind]=sort(z); % sort z in an ascending order
n=size(A,2); % the dimensionality of the data
k=size(y,2); % the number of classes (tasks)
X=zeros(n,k,z_num); % set the size of output X
C=zeros(z_num, k); % set the size of output C
% run the code to compute the first solution
[x, c, funVal]=mcLogisticC(A, y, z_value(1), opts);
X(:,:,z_ind(1))=x; % store the solution x
C(z_ind(1), :)=c; % store the solution c
% set .init for warm start
opts.init=1; % using the defined ones
for i=2:z_num
opts.x0=x; % warm-start of x
opts.c0=c; % warm-start of c
% run the function mcLogisticC
[x, c, funVal]=mcLogisticC(A, y, z_value(i), opts);
X(:,:, z_ind(i))=x; % store the solution x
C(z_ind(i), :)=c; % store the solution c
end
otherwise
fprintf('\n The function value specified in opts.fName is not supported!');
end
Datasets
Models
