Table 8.
Model complexity and efficiency comparison.
| Model | Params (M) | Train (h) | Latency (ms) |
|---|---|---|---|
| CNN ensemble baseline | 58.0 | 4.5 | 12.0 |
| ViT ensemble baseline | 38.0 | 4.0 | 14.0 |
| NeuroTrustNet (AAS) | 39.0 | 0.7 | 15.0 |
| AdaptoVision | 4.2 | 1.2 | 4.0 |
| RapidNet | 3.8 | 1.0 | 3.5 |
Latency is reported per image in milliseconds. Bold values highlight the proposed NeuroTrustNet/AAS configuration.