-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathresults_tmp.txt
63 lines (57 loc) · 3.3 KB
/
results_tmp.txt
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
first results:
------------------
the left value is the test accuracy on the columbia-test set, the right value the cvaccuracy (5-fold cv) on the respective training set.
We can see:
- best model (by cvacc):
x GTZAN: simple_cnn on spectro (0.95323076248168948)
x columbia-train: simple_cnn-linvar on spectro(1.0) # (TODO: too good, check again)
(baseline is: svm with MIRpreprocessing: ~89% # ( TODO: check again)
(# TODO: add test acc and mybe F1 score (if time))
we also want to compare:
- performance of different preprocessings for simple models (linear)
- performance for different preprocessings (more complex models)
----------------------------------------------------------------------------------------------------------------
---------------- Experiment for linear on RhythmData(GTZAN)
(0.80000000000000004, 0.69353845119476321)
---------------- Experiment for linear-linvar on RhythmData(GTZAN)
(0.92500000000000004, 0.85938461538461541)
---------------- Experiment for linear on MIRData(GTZAN)
(0.80000000000000004, 0.53046153783798222)
---------------- Experiment for linear-linvar on MIRData(GTZAN)
(0.47499999999999998, 0.63138461538461532)
---------------- Experiment for linear on RhythmData(columbia-train)
(0.94999999999999996, 0.72499999999999998)
---------------- Experiment for linear-linvar on RhythmData(columbia-train)
(0.94999999999999996, 0.84166666666666679)
---------------- Experiment for linear on MIRData(columbia-train)
(0.5, 0.50833334326744084)
---------------- Experiment for linear-linvar on MIRData(columbia-train)
(0.5, 0.60833333333333317)
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
---------------- Experiment for simple_cnn on SpectroData(columbia-train)
(0.94999999999999996, 0.98333333730697636)
---------------- Experiment for simple_cnn-linvar on SpectroData(columbia-train)
(0.94999999999999996, 0.99166666666666681)
---------------- Experiment for simple_cnn on SpectroData(GTZAN)
(0.94999999999999996, 0.95323076248168948)
---------------- Experiment for simple_cnn-linvar on SpectroData(GTZAN)
(0.90000000000000002, 0.93753846153846143)
---------------- Experiment for simple_cnn-linvar on SpectroData(columbia-train)
(0.97499999999999998, 1.0)
---------------- Experiment for simple_cnn on RhythmData(columbia-train)
(0.94999999999999996, 0.94999999999999996)
---------------- Experiment for simple_cnn-linvar on RhythmData(columbia-train)
(0.97499999999999998, 0.95833333333333326)
---------------- Experiment for simple_cnn on MIRData(columbia-train)
(0.90000000000000002, 0.85833334326744082)
---------------- Experiment for simple_cnn-linvar on MIRData(columbia-train)
(0.94999999999999996, 0.97499999999999998)
---------------- Experiment for simple_cnn on RhythmData(GTZAN)
(0.97499999999999998, 0.90615384578704838)
---------------- Experiment for simple_cnn-linvar on RhythmData(GTZAN)
(0.97499999999999998, 0.93076923076923068)
---------------- Experiment for simple_cnn on MIRData(GTZAN)
(0.94999999999999996, 0.67969231009483333)
---------------- Experiment for simple_cnn-linvar on MIRData(GTZAN)
(0.94999999999999996, 0.94492307692307698)
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX