Table 2.
In-Distribution Classification Results.
| Accuracy % (Confidence Interval) | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| MHEALTH Segment Length |
SPARS Segment Length |
SPARS9x Segment Length |
|||||||||||||
| Method | 2.0s | 4.0s | 6. 0s | 8.0s | 10.0s | 2.0s | 4.0s | 6. 0s | 8.0s | 10.0s | 2.0s | 4.0s | 6. 0s | 8.0s | 10.0s |
| KNN | 88.0 (3.0) | 87.7 (2.3) | 86.5 (2.2) | 87.7 (2.4) | 87.1 (1.5) | 81.0 (1.7) | 82.3 (2.6) | 82.9 (0.83) | 81.4 (2.8) | 82.2 (1.8) | 96.4 (0.67) | 97.6 (0.78) | 97.8 (0.71) | 97.7 (0.80) | 97.7 (0.77) |
| KNN (CRNN Deep Features) | 95.2 (1.4) | 94.1 (2.3) | 94.8 (1.3) | 96.0 (1.1) | 92.7 (1.4) | 87.7 (0.86) | 67.5 (4.7) | 81.2 (1.0) | 78.7 (1.7) | 75.6 (1.2) | 97.6 (0.41) | 96.6 (0.76) | 96.8 (0.92) | 95.5 (0.57) | 94.8 (0.61) |
| KNN (FCN Deep Features) | 93.1 (0.8) | 93.7 (2.3) | 93.3 (1.7) | 92.8 (1.7) | 93.1 (1.8) | 89.9 (0.98) | 92.5 (1.2) | 92.2 (1.1) | 93.1 (1.4) | 92.1 (1.5) | 98.9 (0.19) | 99.8 (0.12) | 99.7 (0.14) | 99.9 (0.066) | 99.4 (0.43) |
| CRNN | 94.8 (1.3) | 93.8 (2.3) | 94.6 (1.4) | 95.9 (1.2) | 90.0 (2.2) | 87.6 (0.87) | 67.4 (5.1) | 80.8 (1.3) | 76.2 (1.7) | 72.9 (0.76) | 97.7 (0.46) | 96.6 (0.76) | 96.7 (0.94) | 95.4 (0.53) | 94.7 (0.64) |
| FCN | 94.2 (1.6) | 94.8 (1.6) | 93.0 (0.79) | 93.5 (1.4) | 95.3 (1.3) | 87.9 (1.4) | 86.4 (1.6) | 87.1 (1.6) | 87.9 (1.2) | 88.0 (2.4) | 98.5 (0.38) | 99.7 (0.19) | 99.6 (0.20) | 99.9 (0.066) | 98.1 (1.7) |