Table 3 Comparison of various proposed VQEDTL models with ISONET.

Classical layer

17

6

6

6

6

Quantum layer

–

4

4

4

4

Qubit

–

4

4

4

4

Train parameter

3,278,371

18,337

16,289

15,265

17,825

Training time

40 min

74 min

130.5 m

46 min

123.5 min

Training loss

0.5505

0.0546

0.2793

0.0987

0.0813

Validation loss

0.5358

0.0437

0.2333

0.1548

0.0892

Test loss

0.5138

0.0247

0.2573

0.1413

0.0754

Train accuracy

0.9404

0.9837

0.8856

0.9710

0.9763

Valid accuracy

0.9468

0.9908

0.8890

0.9532

0.9752

Test accuracy

0.9460

0.9926

0.8827

0.9609

0.9814

Precision

0.94

0.99

0.87

0.96

0.98

Recall

0.93

0.99

0.88

0.96

0.98

F1score

0.93

0.99

0.87

0.96

0.98