-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathV9_LocalKernelCalculation.m
146 lines (125 loc) · 4.12 KB
/
V9_LocalKernelCalculation.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
function [KHL] = V9_LocalKernelCalculation(KH , NNRate, cla_num)
ker_num = size(KH, 3);
smp_num = size(KH, 1);
gamma0 = ones(ker_num,1)/ker_num;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Generate average kernel
KH_temp = KH;
% remove too large components
KH_temp = remove_large(KH_temp); % 设定上界下界
% base kernel normalization
KH_temp = knorm(KH_temp);
% base kernel centerlization
KH_temp = kcenter(KH_temp);
% divide base kernels with std
KH_temp = divide_std(KH_temp);
% Generate average kernel
AVGKer = mycombFun(KH_temp,gamma0);
%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Generate affinity matrix
Org_AVGKer = AVGKer;
NeiNum = round( NNRate * smp_num / cla_num );
[aff_matrix, indx] = genarateNeighborhood(AVGKer , NeiNum);
aff_matrix = (aff_matrix + aff_matrix');
% Cut index that are not good
% aff_matrix = cut_link(aff_matrix, indx, NeiNum, cla_num);
%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% filter the extreme values in each single kernel
KHL = zeros(size(KH));
for i_ker = 1 : size(KH, 3)
Current_kernel = KH(:,:,i_ker) .* aff_matrix;
Current_kernel(Current_kernel < 0) = 0;
mean_value = mean(Current_kernel(:));
std_value = std(Current_kernel(:));
up_threshold = mean_value + std_value*3;
down_threshold = max(0, mean_value - std_value*3);
Current_kernel(Current_kernel > up_threshold) = up_threshold; %上界 下界
Current_kernel(Current_kernel < down_threshold) = down_threshold;
Current_kernel = kernel_completion(Current_kernel, Org_AVGKer);
KHL(:,:,i_ker) = Current_kernel;
end
end
%% Generate neighbor index
function [aff_matrix, indx] = genarateNeighborhood(KC,tau)
aff_matrix = zeros(size(KC, 1));
smp_num = size(KC, 1);
KC0 = KC - 10^18*eye(smp_num);
% KC0 = KC;
[~,indx] = sort(KC0,'descend');
indx_0 = indx(1:tau,:);
smp_index = 1 : smp_num;
smp_index = repmat(smp_index, tau, 1);
indx_0 = indx_0(:);
smp_index = smp_index(:);
real_index = (smp_index-1) * smp_num + indx_0;
aff_matrix(real_index) = 1;
end
function aff_matrix = cut_link(aff_matrix, indx, NeiNum, cla_num)
similarity_sum = sum(aff_matrix, 2);
smp_num = size(aff_matrix, 1);
smp_threshold = round(smp_num/cla_num*1.2);
smp_threshold = min(smp_threshold, 4*NeiNum);
index = find(similarity_sum > smp_threshold);
index2 = indx(:, index);
index2 = index2(1:round(NeiNum/2), :);
aff_matrix(index, :) = 0;
aff_matrix(:, index) = 0;
index = repmat(index', round(NeiNum/2), 1);
index = index(:);
index2 = index2(:);
final_index = (index-1)*smp_num + index2;
temp_zero = zeros(smp_num);
temp_zero(final_index) = 1;
temp_zero = (temp_zero + temp_zero')>0;
aff_matrix = double(aff_matrix);
aff_matrix = aff_matrix + temp_zero;
end
function kernel = kernel_completion(kernel, org_kernel)
Avg_index = kernel ~= 0;
Ker_sum = sum(Avg_index, 2);
index = find(Ker_sum == 0);
threshold = mean(kernel(:));
if ~isempty(index)
org_kernel = org_kernel - diag(diag(org_kernel));
Small_samples = org_kernel(index, :);
[~, smp_indexes] = sort(Small_samples,2,'descend');
smp_indexes = smp_indexes(:,1:2);
for ii = 1 : size(Small_samples, 1)
kernel(index(ii), smp_indexes(ii,:)) = threshold;
end
end
end
function K = remove_large(K)
if size(K,3)>1
for i=1:size(K,3)
current_kernel = K(:,:,i);
mean_value = mean(current_kernel(:));
std_value = std(current_kernel(:));
threshold_up = mean_value + std_value*4; %上 界 下界
threshold_dn = mean_value - std_value*4;
current_kernel(current_kernel > threshold_up) = threshold_up;
current_kernel(current_kernel < threshold_dn) = threshold_dn;
K(:,:,i) = current_kernel;
end
else
mean_value = mean(K(:));
std_value = std(K(:));
threshold_up = mean_value + std_value*4;
threshold_dn = mean_value - std_value*4;
K(K > threshold_up) = threshold_up;
K(K < threshold_dn) = threshold_dn;
end
end
function K = divide_std(K)
if size(K,3)>1
for i=1:size(K,3)
current_kernel = K(:,:,i);
std_value = std(current_kernel(:));
current_kernel = current_kernel / std_value;
K(:,:,i) = current_kernel;
end
else
std_value = std(K(:));
K = K / std_value;
end
end