How can I plot the Cumulative Match Score Characteristic curve for my output CNN?

4 Ansichten (letzte 30 Tage)
Alaa
Alaa am 17 Mär. 2016
Bearbeitet: Alaa am 17 Mär. 2016
I use the CNN for identification task and I found that all the papers try using the Cumulative Match Score Characteristic (CMC) curve to visualize the results. I understand what the meaning of the CMC curve, but I don't know how plot it in my case. I use the Softmax on the top of my CNN and I calculate the accuracy of my model on the testing dataset after I get the output matrix from the Softmax as follows:
[max_val, estimated_labels] = max(outputs);
estimated_labels = estimated_labels';
correct_count = length(find(estimated_labels == test_labels));
acc = correct_count / size(test_imgs, 4);
fprintf('Overall test accuracy: %f%%\n', acc * 100);
Any help in this regard it highly appreciated.
Here is some values not all from each matrix in my code:
max_val=0.7812
0.9998
0.9808
0.9985
0.9993
0.9962
0.5600
0.9986
0.9675
0.9755
1.0000
0.9917
0.6297
0.9948
0.9950
0.4407
0.9727
0.9864
1.0000
0.9998
0.8779
0.9999
0.8765
0.4498
0.7942
0.9905
0.5248
0.9999
0.9912
1.0000
0.3731
0.8184
1.0000
0.9759
0.9692
0.8997
0.8519
1.0000
0.4999
0.8744
0.9714
0.7590
1.0000
0.6382
0.9998
0.8760
0.9825
0.9970
0.8468
0.9821
1.0000
0.9847
1.0000
0.4963
1.0000
1.0000
0.9947
estimated_labels=
98
26
44
20
75
54
48
35
71
60
61
66
21
69
73
86
88
23
29
29
41
89
12
45
67
32
52
56
67
10
6
64
27
60
76
69
8
35
22
5
87
71
96
90
75
52
22
76
36
88
28
74
30
16
106
53
93
test_labels=
98
26
44
20
75
54
48
35
71
60
61
66
21
69
73
86
88
23
29
29
41
89
12
45
67
32
52
56
67
10
6
64
27
60
76
69
8
35
22
5
87
71
96
94
75
52
22
76
36
88
28
74
30
16
106
53
93

Antworten (0)

Kategorien

Mehr zu Deep Learning Toolbox finden Sie in Help Center und File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by