Research Article
Kernel Negative ε Dragging Linear Regression for Pattern Classification
Table 1
Accuracies (%) of different methods on the GT database.
| Number of training samples per class | 5 | 6 | 7 | 8 | 9 | 10 |
| Our method | 70.80 | 73.53 | 76.05 | 80.26 | 81.33 | 82.36 | DLSR | 50.48 | 50.93 | 52.80 | 53.77 | 53.77 | 54.56 | CLSR | 63.58 | 65.33 | 67.30 | 70.57 | 71.00 | 71.40 | NNLS | 68.90 | 71.02 | 73.88 | 76.83 | 78.27 | 79.68 | -SVM | 70.42 | 72.20 | 74.93 | 79.46 | 80.17 | 81.92 | KNN | 59.30 | 61.76 | 63.85 | 67.49 | 69.23 | 69.40 | SRC | 56.18 | 58.44 | 60.13 | 63.09 | 63.57 | 64.36 | LRC | 68.90 | 71.22 | 73.78 | 77.71 | 79.20 | 80.68 | NDLR | 65.66 | 66.76 | 69.38 | 73.60 | 73.57 | 74.56 |
|
|