/
runReal.m
executable file
·166 lines (143 loc) · 4.28 KB
/
runReal.m
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
function res = runReal(X, Y, weight_func, k, max_it, corr_thres1, corr_thres2)
% Function to run real data
% Output - res, cells store results
% Note: default to use correlation graph now
% Methods to run in this script
run_Lasso = 0;
run_MTLasso = 0;
run_MTLasso_LG = 0;
run_MTLasso_FG = 0;
run_MTLasso_2G = 1;
run_correlation_method = 1;
N = size(X, 1);
% Normalization
X = X - repmat(mean(X), N, 1);
Y = Y - repmat(mean(Y), N, 1);
% Inits and graphs
G1.C = tril(nanFilter(corr(Y)), -1);
inds1 = find(abs(G1.C) > corr_thres1);
G1.E = inds2subs(inds1, size(G1.C)) - 1;
G1.C = G1.C(inds1);
G2.C = tril(nanFilter(corr(X)), -1);
inds2 = find(abs(G2.C) > corr_thres2);
G2.E = inds2subs(inds2, size(G2.C)) - 1;
G2.C = G2.C(inds2);
switch weight_func
case 'abs'
G1.W = abs(G1.C);
G2.W = abs(G2.C);
case 'minus' % can only be used for correlation graph
G1.W = abs(G1.C) - corr_thres1;
G2.W = abs(G2.C) - corr_thres2;
case 'minus_square'
G1.W = (abs(G1.C) - corr_thres1) .^ 2;
G2.W = (abs(G2.C) - corr_thres2) .^ 2;
end
init_val.B = pinv(X) * Y;
% Inits for CV
block_size = floor(N / k);
for i = 1:k
test_id = (i - 1) * block_size + 1 : min(i * block_size, N);
training_id = setdiff(1:N, test_id);
training_X = X(training_id, :);
training_Y = Y(training_id, :);
init_val.Bk{k} = pinv(training_X) * training_Y;
end
%% Run standard lasso
% Run Cross-validation gradient descent to choose hyperparameters
if run_Lasso == 1
lambda_init = 100;
gamma1_init = 0;
gamma2_init = 0;
tic
[lambda_0, gamma1_0, gamma2_0, CV_trace_0] = crossValidation(X, Y, G1, G2,...
k, max_it, lambda_init, gamma1_init, gamma2_init, init_val);
fprintf('1 Cross validation done...\n');
toc
% Main function
tic
Bhat_0 = MTLasso_2graph(X, Y, G1, G2, lambda_0, gamma1_0, gamma2_0, init_val);
fprintf('1 Main program done...\n');
toc
end
%% Run standard multi-task lasso
if run_MTLasso == 1
lambda_init = 100;
tic
[lambda_MT CV_trace_MT] = crossValidation_MTLasso(X, Y, k, max_it, lambda_init, init_val);
fprintf('2 Cross Validation done...\n');
toc
% Main function
tic
Bhat_MT = MTLasso(X, Y, lambda_MT, init_val);
fprintf('2 Main program done...\n');
toc
end
%% Run one-graph guided multi-task lasso
% Run Cross-validation gradient descent to choose hyperparameters
if run_MTLasso_LG == 1
lambda_init = 100;
gamma1_init = 10;
gamma2_init = 0;
tic
[lambda_1, gamma1_1, gamma2_1, CV_trace_1] = crossValidation(X, Y, G1, G2,...
k, max_it, lambda_init, gamma1_init, gamma2_init, init_val);
fprintf('3 Cross validation done...\n');
toc
% Main function
tic
Bhat_1 = MTLasso_2graph(X, Y, G1, G2, lambda_1, gamma1_1, gamma2_1, init_val);
fprintf('3 Main program done...\n');
toc
end
%% Run one-graph guided multi-task lasso, here the graph is on features
% Run Cross-validation gradient descent to choose hyperparameters
if run_MTLasso_FG == 1
lambda_init = 100;
gamma1_init = 0;
gamma2_init = 10;
tic
[lambda_1_prime, gamma1_1_prime, gamma2_1_prime, CV_trace_1_prime] = crossValidation(X, Y, G1, G2,...
k, max_it, lambda_init, gamma1_init, gamma2_init, init_val);
fprintf('4 Cross validation done...\n');
toc
% Main function
tic
Bhat_1_prime = MTLasso_2graph(X, Y, G1, G2, lambda_1_prime, gamma1_1_prime, gamma2_1_prime, init_val);
fprintf('4 Main program done...\n');
toc
end
%% Run two-graph guided multi-task lasso
if run_MTLasso_2G == 1
lambda_init = 100;
gamma1_init = 10;
gamma2_init = 10;
tic
[lambda_2, gamma1_2, gamma2_2, CV_trace_2] = crossValidation(X, Y, G1, G2,...
k, max_it, lambda_init, gamma1_init, gamma2_init, init_val);
toc
% Main function
tic
Bhat_2 = MTLasso_2graph(X, Y, G1, G2, lambda_2, gamma1_2, gamma2_2, init_val);
fprintf('Main program done...\n');
toc
end
%% Run correlation method
if run_correlation_method == 1
% TBI
end
% Save values
res = cell(12, 1);
% res{1} = Bhat_0;
% res{2} = Bhat_MT;
% res{3} = Bhat_1;
% res{4} = Bhat_1_prime;
res{5} = Bhat_2;
% res{6} = CV_trace_0;
% res{7} = CV_trace_MT;
% res{8} = CV_trace_1;
% res{9} = CV_trace_1_prime;
res{10} = CV_trace_2;
res{11} = G1;
res{12} = G2;
save('real_result', 'res');d