Table 3. Mathematical framework for model training parameters based on database size.
| Net
Transfer |
OBS | TS
(OBS x 70%) |
VS
(OBS x 30%) |
MBS | IPE
(TS/MBS) |
EPOCHS | MI
(EPOCH X IPE) |
|---|---|---|---|---|---|---|---|
| I | 4854 | 3398 ≈ (4854 x 0.7) | 1456 ≈ (4854 x 0.3) | 10 | 340 ≈ (3398/10) | 6 | 2040 = (6 x 340) |
| II | 5255 | 3678 ≈ (5255 x 0.7) | 1577 ≈ (5255 x 0.3) | 10 | 368 ≈ (3678/10) | 6 | 2208 = (6 x 368) |
| III | 5300 | 3710 ≈ (5300 x 0.7) | 1590 ≈ (5300 x 0.3) | 10 | 371 ≈ (3710/10) | 6 | 2226 = (6 x 371) |
| IV | 6787 | 4751 ≈ (6787 x 0.7) | 2036 ≈ (6787 x 0.3) | 10 | 475 ≈ (4751/10) | 6 | 2850 = (6 x 368) |
| V | 9860 | 6902 ≈ (9860 x 0.7) | 2958 ≈ (9860 x 0.3) | 10 | 690 ≈ (6902/10) | 6 | 4140 = (6 x 690) |
Glossary: Observations (OBS), Test Set (TS), Validation Set (VS), Mini Batch Size (MBS), Iteration Per Epoch (IPE), Maximum Iterations (MI).