Path Loss Prediction on Indoor Environment Path Loss Dataset
5.14MAETabNet
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| TabNetExpression=Feature importance: nw: 0.1332, nf: 0.2643, d: 0.5505, f: 0.0520, Interpretability=Low, Validity=✗2025.05 | 5.14 | 44.02 | 5.09 | 82 | |
| ResNet-MLPExpression=/, Interpretability=None, Validity=✗2025.05 | 5.67 | 54.97 | 5.54 | 78 | |
| KAN [4,1]Expression=Learned tree and splines as in Figure 11, Interpretability=Medium, Validity=?2025.05 | 5.89 | 57.67 | 5.84 | 77 | |
| KAN [4,1] Auto-symbolicExpression=41.5 log10 (0.4d + 4.03) − 1.3 cos (5.04f − 0.5) + 4.9 sin (8.2nw − 2.9) + 52.03 log10 (2.4nf + 8.6) − 11.2, Interpretability=High, Validity=✗2025.05 | 6.33 | 63.51 | 6.33 | 74 | |
| DSR-VPGExpression=nf^2 + 4nf + nfnw + 80, Interpretability=High, Validity=✗2025.05 | 6.39 | 64.25 | 6.48 | 75 | |
| DSR-PQTExpression=d/2 + 4nw + 80, Interpretability=High, Validity=✗2025.05 | 6.48 | 66.11 | 6.59 | 74 | |
| DSR-RSPGExpression=d/2 + 3 log10(f) + 5nw + 70, Interpretability=High, Validity=✓2025.05 | 6.58 | 67.85 | 6.69 | 73 | |
| PL^{EI}Expression=10n log10(d) + PL0 + nwLw + ((nf + 2) / (nf + 1) - b)^nf Lf, Interpretability=High, Validity=✓2025.05 | 7.64 | 111.67 | 7.52 | 49 | |
| PL^{MWF}Expression=10n log10 (d) + PL0 + WAF + FAF, Interpretability=High, Validity=✓2025.05 | 9.01 | 121.54 | 8.98 | 56 |