Research Article
A Constrained Algorithm Based NMFα for Image Representation
Table 5
The comparison of classification accuracy rates on the Caltech 101 Database.
| | Ac() (%) | CNMF | | | | |
| 2 | 66.39 | 78.80 | 78.31 ± 1.69 | 80.10 ± 2.07 | 82.76 ± 3.56 | 3 | 60.91 | 69.98 | 68.77 ± 1.57 | 72.94 ± 1.51 | 73.59 ± 3.51 | 4 | 50.82 | 57.50 | 58.19 ± 1.98 | 56.28 ± 1.96 | 59.89 ± 2.82 | 5 | 50.83 | 53.67 | 53.71 ± 1.02 | 54.12 ± 1.44 | 55.06 ± 2.63 | 6 | 50.49 | 55.05 | 55.63 ± 0.89 | 58.04 ± 1.47 | 58.97 ± 1.97 | 7 | 46.09 | 51.12 | 50.04 ± 0.11 | 51.73 ± 0.25 | 52.36 ± 1.81 | 8 | 44.21 | 50.50 | 51.72 ± 0.51 | 48.37 ± 1.59 | 52.64 ± 1.52 | 9 | 41.34 | 46.58 | 47.85 ± 0.28 | 45.87 ± 0.30 | 49.54 ± 1.23 | 10 | 40.80 | 46.02 | 46.02 ± 0.05 | 45.76 ± 0.08 | 46.02 ± 0.26 |
| Avg. | 50.21 | 56.58 | 56.69 | 57.02 | 58.98 |
|
|