-
Notifications
You must be signed in to change notification settings - Fork 36
/
Copy pathRunClustering.m
118 lines (90 loc) · 7.39 KB
/
RunClustering.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
function RunClustering(DataSetStartIndex, DataSetEndIndex, ClusteringMethod, Method, RepType)
Methods = [cellstr('Random'), 'KShape'];
Types = [cellstr('ZExact'), 'Z5', 'Z10', 'Z20', 'Z98per', 'Z95per', 'Z90per', 'Z85per', 'Z80per'];
% first 2 values are '.' and '..' - UCR Archive 2018 version has 128 datasets
dir_struct = dir('/rigel/dsi/users/ikp2103/VLDBGRAIL/UCR2018/');
Datasets = {dir_struct(3:130).name};
% Sort Datasets
[Datasets, DSOrder] = sort(Datasets);
Results = zeros(length(Datasets),2);
for i = 1:length(Datasets)
if (i>=DataSetStartIndex & i<=DataSetEndIndex)
display(['Dataset being processed: ', char(Datasets(i))]);
DS = LoadUCRdataset(char(Datasets(i)));
% Get Kernel Matrix
%KM = dlmread( strcat( 'DATASETS/',char(Datasets(i)),'/', char(Datasets(i)), '_CCKernel_', num2str(gamma) ,'.kernelmatrix'));
%KM = dlmread( strcat( 'DATASETS/',char(Datasets(i)),'/', char(Datasets(i)), '_NCCc', '.distmatrix'));
%KM = DM2Kernel(KM);
for rep = 1 : 1
rep
rng(rep);
% Find appropriate gamma for this rep for this
% dataset
TestVarianceNew = dlmread( strcat( 'RunTestVarianceApproximate/', 'RESULTS_RunTestVarianceApproximate_', char(Datasets(i)), '_', char(Methods(Method)), '_',num2str(rep) ,'.Results'));
gamma = TestVarianceNew(1);
% Extract Sample Points
ZExact = dlmread( strcat( 'REPRESENTATIONSGamma', num2str(gamma),'/',char(Datasets(i)),'/','RepLearningFixedSamples', '_', char(Methods(Method)), '_', num2str(rep) ,'.Zexact') );
Z5 = dlmread( strcat( 'REPRESENTATIONSGamma', num2str(gamma),'/',char(Datasets(i)),'/','RepLearningFixedSamples', '_', char(Methods(Method)), '_', num2str(rep) ,'.Ztop5') );
Z10 = dlmread( strcat( 'REPRESENTATIONSGamma', num2str(gamma),'/',char(Datasets(i)),'/','RepLearningFixedSamples', '_', char(Methods(Method)), '_', num2str(rep) ,'.Ztop10') );
Z20 = dlmread( strcat( 'REPRESENTATIONSGamma', num2str(gamma),'/',char(Datasets(i)),'/','RepLearningFixedSamples', '_', char(Methods(Method)), '_', num2str(rep) ,'.Ztop20') );
Z98per = dlmread( strcat( 'REPRESENTATIONSGamma', num2str(gamma),'/',char(Datasets(i)),'/','RepLearningFixedSamples', '_', char(Methods(Method)), '_', num2str(rep) ,'.Z98per') );
Z95per = dlmread( strcat( 'REPRESENTATIONSGamma', num2str(gamma),'/',char(Datasets(i)),'/','RepLearningFixedSamples', '_', char(Methods(Method)), '_', num2str(rep) ,'.Z95per') );
Z90per = dlmread( strcat( 'REPRESENTATIONSGamma', num2str(gamma),'/',char(Datasets(i)),'/','RepLearningFixedSamples', '_', char(Methods(Method)), '_', num2str(rep) ,'.Z90per') );
Z85per = dlmread( strcat( 'REPRESENTATIONSGamma', num2str(gamma),'/',char(Datasets(i)),'/','RepLearningFixedSamples', '_', char(Methods(Method)), '_', num2str(rep) ,'.Z85per') );
Z80per = dlmread( strcat( 'REPRESENTATIONSGamma', num2str(gamma),'/',char(Datasets(i)),'/','RepLearningFixedSamples', '_', char(Methods(Method)), '_', num2str(rep) ,'.Z80per') );
%ZResults = dlmread( strcat( 'APPROXREPRESENTATIONS100/',char(Datasets(i)),'/','RepLearningTuned', '_', char(Methods(Method)), '_', num2str(rep) ,'.Results'));
tic;
if ClusteringMethod == 1
if RepType == 1
mem = KernelKmeansClustering(ZExact, ZExact, length(DS.ClassNames));
elseif RepType == 2
mem = KernelKmeansClustering(ZExact, Z5, length(DS.ClassNames));
elseif RepType == 3
mem = KernelKmeansClustering(ZExact, Z10, length(DS.ClassNames));
elseif RepType == 4
mem = KernelKmeansClustering(ZExact, Z20, length(DS.ClassNames));
elseif RepType == 5
mem = KernelKmeansClustering(ZExact, Z98per, length(DS.ClassNames));
elseif RepType == 6
mem = KernelKmeansClustering(ZExact, Z95per, length(DS.ClassNames));
elseif RepType == 7
mem = KernelKmeansClustering(ZExact, Z90per, length(DS.ClassNames));
elseif RepType == 8
mem = KernelKmeansClustering(ZExact, Z85per, length(DS.ClassNames));
elseif RepType == 9
mem = KernelKmeansClustering(ZExact, Z80per, length(DS.ClassNames));
end
elseif ClusteringMethod == 2
if RepType == 1
mem = KernelSCApprox(ZExact, length(DS.ClassNames));
elseif RepType == 2
mem = KernelSCApprox(Z5, length(DS.ClassNames));
elseif RepType == 3
mem = KernelSCApprox(Z10, length(DS.ClassNames));
elseif RepType == 4
mem = KernelSCApprox(Z20, length(DS.ClassNames));
elseif RepType == 5
mem = KernelSCApprox(Z98per, length(DS.ClassNames));
elseif RepType == 6
mem = KernelSCApprox(Z95per, length(DS.ClassNames));
elseif RepType == 7
mem = KernelSCApprox(Z90per, length(DS.ClassNames));
elseif RepType == 8
mem = KernelSCApprox(Z85per, length(DS.ClassNames));
elseif RepType == 9
mem = KernelSCApprox(Z80per, length(DS.ClassNames));
end
end
ClusteringTime = toc;
RI = RandIndex(mem, DS.DataClassLabels);
% Evaluate SmplPoints in terms of clustering
% measures (e.g., SSE, RandIndex, NystromAppx)
ResultsTmp = [RI,ClusteringTime];
%
Results(i,:) = Results(i,:) + ResultsTmp;
end
Results(i,:) = Results(i,:) ./ 1;
end
dlmwrite( strcat( 'RunClustering/','RESULTS_RunClustering_', num2str(ClusteringMethod), '_', char(Methods(Method)), '_', char(Types(RepType)), '_', num2str(DataSetStartIndex), '_', num2str(DataSetEndIndex),'.results'), Results, 'delimiter', '\t');
end
end