16、稀疏支持向量机：原理、方法与性能评估

稀疏支持向量机：原理、方法与性能评估

1. 线性规划支持向量机性能对比

线性规划支持向量机（LP SVM）有一对多（One-against-all）和成对（Pairwise）两种类型。以下是不同数据集上这两种类型LP SVM的性能对比：
| 数据 | 项目 | 一对多LP SVM | 成对LP SVM |
| — | — | — | — |
| Iris | Parm | γ0.1, C100 | γ1, C5,000 |
| | Dis. | 92.00 (94.67) | 92.00 |
| | Min | 94.67 (97.33) | 92.00 |
| | Avg. | 94.67 (97.33) | 92.00 |
| | SVs | 3 | 3 |
| Numeral | Parm | d2, C1 | γ10, C5,000 |
| | Dis. | 99.27 (99.63) | 99.39 |
| | Min | 99.51 (99.75) | 99.39 |
| | Avg. | 99.51 (99.75) | 99.39 |
| | SVs | 7 | 4 |
| Thyroid | Parm | d2, C5,000 | Linear, C5,000 |
| | Dis. | 96.00 (99.28) | 97.35 (98.59) |
| | Min | 97.61 (99.58) | 97.64 (98.78) |
| | Avg. | 97.61 (99.58) | 97.67 (98.78) |
| | SVs |