Table 3 Hyperparameter settings for investigated models.
From: Effective deep learning aided vehicle classification approach using Seismic Data
Model | SVM | CNN | LSTM | Proposed CL |
|---|---|---|---|---|
Batch size | N/A | 32 | 64 | 128 |
Learning rate | N/A | 0.001 | 0.0005 | 0.0003 |
Optimizer | N/A | Adam | RMSprop | AdamW |
Loss function | Hinge loss | Cross-entropy | Cross-entropy | InfoNCE loss |
Epochs | N/A | 100 | 120 | 150 |
Dropout | N/A | 0.5 | 0.3 | 0.2 |
Regularization | L2 (C = 1.0) | L2 (0.0001) | L2 (0.0005) | Weight decay (0.001) |
