Research Article
The Optimisation for Local Coupled Extreme Learning Machine Using Differential Evolution
Table 6
Classification results of ELM, OSFuzzyELM, LC-ELM, and ELC-ELM.
| Dataset | Algorithm | Training (%) | Training time | Testing (%) | Testing time | Number of nodes/rules | Accy | SD | Accy | SD |
| Ecoli | ELM | 89.78 | 0.90 | 0.0062 | 85.71 | 1.68 | 0.0031 | 20 | OSFuzzyELM | 90.04 | 1.53 | 0.1045 | 86.79 | 2.72 | 0.0047 | 5 | LC-ELM | 89.29 | 1.17 | 0.0218 | 87.14 | 4.23 | 0.0156 | 20 | ELC-ELM | 90.89 | 0.74 | 312.5917 | 89.29 | 4.29 | 0.0162 | 20 |
| Glass | ELM | 72.25 | 3.82 | 0.0047 | 63.47 | 5.36 | 0.0016 | 20 | OSFuzzyELM | 92.04 | 1.88 | 0.1529 | 63.33 | 4.05 | 0.0062 | 10 | LC-ELM | 74.79 | 2.97 | 0.0172 | 63.75 | 6.56 | 0.0062 | 20 | ELC-ELM | 73.59 | 3.31 | 21.9727 | 66.39 | 6.98 | 0.0074 | 20 |
| Ionosphere | ELM | 91.70 | 1.69 | 0.0078 | 83.51 | 3.12 | 0.0031 | 40 | OSFuzzyELM | 95.88 | 3.29 | 0.1357 | 80.91 | 4.98 | 0.0031 | 3 | LC-ELM | 93.86 | 2.05 | 0.0359 | 86.49 | 2.68 | 0.0218 | 40 | ELC-ELM | 93.53 | 1.70 | 121.4764 | 86.84 | 3.51 | 0.0236 | 40 |
| Iris | ELM | 98.30 | 0.67 | 0.0031 | 96.60 | 2.50 | 0.0016 | 15 | OSFuzzyELM | 98.50 | 0.71 | 0.0296 | 96.00 | 2.11 | 0.0031 | 5 | LC-ELM | 97.90 | 1.29 | 0.0078 | 96.40 | 2.07 | 0.0047 | 15 | ELC-ELM | 97.40 | 0.52 | 11.8374 | 97.20 | 2.70 | 0.0054 | 15 |
| Sonar | ELM | 88.70 | 2.92 | 0.0031 | 75.57 | 3.95 | 0.0016 | 40 | OSFuzzyELM | 97.46 | 3.35 | 0.2558 | 67.00 | 8.21 | 0.0062 | 3 | LC-ELM | 89.42 | 4.06 | 0.0343 | 76.29 | 6.25 | 0.0187 | 40 | ELC-ELM | 89.78 | 2.95 | 117.0569 | 78.00 | 6.61 | 0.0178 | 40 |
| Wisconsin | ELM | 97.71 | 0.55 | 0.0047 | 96.36 | 1.42 | 0.0031 | 40 | OSFuzzyELM | 97.54 | 0.38 | 0.1934 | 96.45 | 1.10 | 0.0094 | 5 | LC-ELM | 97.49 | 0.43 | 0.0874 | 96.97 | 1.06 | 0.0406 | 40 | ELC-ELM | 97.41 | 0.53 | 250.2381 | 97.54 | 1.59 | 0.0397 | 40 |
|
|