Table 6 Summary of accuracy measurements for the proposed 3 deep learning models, the heuristic score function method based on previous work and a simple random generator. The accuracy measurements were performed over 16 datasets that each contain \(N=2500\) randomly generated data groups associated to routing problems set under the same routing and environmental parameters specific to the dataset.
From: Machine learning optimal ordering in global routing problems in semiconductors
Dataset | Accuracy(\(\%\)) | ||||
|---|---|---|---|---|---|
Model 1 | Model 2 | Model 3 | Heuristic | Random | |
Data 1 | 40.69 | 41.30 | 36.63 | Â | 16.67 |
Data 2 | 41.30 | 40.89 | 33.82 | 17.25 | |
Data 3 | 39.72 | 39.11 | 36.90 | Â | |
Data 4 | 40.93 | 40.93 | 38.10 | 15.44 | |
Data 5 | 40.44 | 42.05 | 34.21 | Â | |
Data 6 | 40.04 | 41.25 | 33.20 | 14.41 | |
Data 7 | 36.87 | 37.27 | 32.46 | Â | |
Data 8 | 36.47 | 38.08 | 31.46 | 14.51 | |
Data 9 | 7.22 | 9.69 | 4.12 | Â | 0.83 |
Data 10 | 7.63 | 7.84 | 3.30 | 0.70 | |
Data 11 | 8.81 | 7.79 | 2.66 | Â | |
Data 12 | 8.40 | 8.20 | 2.87 | 0.82 | |
Data 13 | 7.58 | 7.99 | 4.11 | Â | |
Data 14 | 7.38 | 7.58 | 2.87 | 0.25 | |
Data 15 | 7.76 | 8.16 | 4.29 | Â | |
Data 16 | 7.35 | 7.55 | 4.08 | 0.78 | |
