Research Article
Self-Trained LMT for Semisupervised Learning
Table 6
Average rankings of the algorithms in 20% labeled ratio (Friedman) and Holm/Hochberg (alpha = 0.05).
| | Algorithm | Friedman
ranking | value | Holm/
Hochberg test |
| | Self-LMT | 5.1923 | — | — | | Cobagging (C45) | 8.7019 | 5.8533E − 03 | 0.0500 | | Tri-training (C45) | 8.7692 | 4.9736E − 03 | 0.0250 | | Democratic-co | 8.7788 | 4.8582E − 03 | 0.0167 | | Cotraining (C45) | 9.5288 | 6.6111E − 04 | 0.0125 | | Co-Forest | 9.7019 | 3.9842E − 04 | 0.0100 | | Self-training (C45) | 10.0288 | 1.4596E − 04 | 0.0083 | | Cotraining (SMO) | 10.2596 | 6.9191E − 05 | 0.0071 | | Cobagging (SMO) | 10.4519 | 3.6267E − 05 | 0.0063 | | Tri-training (SMO) | 10.8654 | 8.4002E − 06 | 0.0056 | | DE-tri-training (SMO) | 11.1538 | 2.8515E − 06 | 0.0050 | | DE-tri-training (C45) | 11.7404 | 2.7211E − 07 | 0.0045 | | Self-training (SMO) | 12.0192 | 8.2869E − 08 | 0.0042 | | SETRED | 12.0288 | 7.9474E − 08 | 0.0038 | | Self-training (NN) | 12.7019 | 3.7052E − 09 | 0.0036 | | ADE-Co-Forest | 12.8558 | 1.7697E − 09 | 0.0033 | | SNNRCE | 12.9135 | 1.3364E − 09 | 0.0031 | | Tri-training (NN) | 13.1538 | 4.0597E − 10 | 0.0029 | | APSSC | 13.8462 | 1.0806E − 11 | 0.0028 | | CLCC | 15.6058 | 2.9085E − 16 | 0.0026 | | Rel-Rasco (NB) | 15.7019 | 1.5503E − 16 | 0.0025 | | Rasco (C45) | 17.0000 | 1.8292E − 20 | 0.0024 |
|
|
