Table 7 Comparative analysis of models based on complexity, robustness, and performance.

Computational Complexity (GFLOPs)

1.8

3.2

2.9

3.5

3.8

Model Parameters (Million)

5.4M

12.1 M

9.8 M

11.5 M

13.2 M

Inference Time (ms per sample)

4.5 ms

8.3 ms

7.1 ms

9.2 ms

10.5 ms

Energy Consumption (mJ per sample)

6.2 mJ

11.5 mJ

9.8 mJ

12.4 mJ

14.1 mJ

Accuracy Drop (%) under Adversarial Attack

1.80%

5.20%

4.70%

5.50%

6.10%

Encryption Time (ms)

4.5 ms

5.8 ms

5.1 ms

5.9 ms

6.8 ms

Decryption Time (ms)

3.3 ms

5.4 ms

8.2 ms

9.0 ms

6.0 ms

False Positive Rate (FPR%)

0.25%

2.10%

1.80%

2.40%

3.00%

Model Size (MB)

14 MB

25 MB

22 MB

28 MB

32 MB

Scalability (Training Time per Epoch in Seconds)

42 s

58 s

54 s

61 s

72 s