Research Article
Mobile Anomaly Detection Based on Improved Self-Organizing Maps
Table 3
The results on universal datasets.
| | Dataset | Size | Algorithm 1 | Algorithm 2 | Traditional -means | Improved -means | | max AR | min AR | Ave AR | Threshold | AR | | Ave AR | | AR |
| | Aggregation | | 98.73% | 89.24% | 93.48% | 0.771 | 99.37% | 40 | 99.56% | 40 | 99.62% | | Compound | | 87.50% | 78.75% | 83.50% | 0.72 | 93.75% | 49 | 97.64% | 49 | 99.00% | | Flame | | 100.00% | 85.42% | 92.08% | 0.75 | 100.00% | 55 | 98.92% | 55 | 100.00% | | Jain | | 97.33% | 89.33% | 92.13% | 0.89 | 100.00% | 29 | 99.95% | 29 | 100.00% | | Path-based | | 93.33% | 71.67% | 83.33% | 0.51 | 90.00% | 50 | 99.00% | 50 | 100.00% | | Spiral | | 82.54% | 30.15% | 56.03% | 0.51 | 66.67% | 27 | 84.36% | 27 | 100.00% |
|
|