Table 5 The characteristics of several MLP structures.

From: Study of CO2 capture by synthesized composite and modelling with machine learning and response surface methodology

Run

Hidden layer

Train function

R

MSE

Epoch

1

2

q

Net

New data

1

6

4

Trainscg

0.988871

0.0032209

0.015980

14

2

6

5

Trainscg

0.997546

0.0014662

0.025214

25

3

6

5

Trainscg

0.996551

0.0014323

0.025667

16

4

10

10

Trainscg

0.985651

0.0027722

0.025556

22

5

6

14

Trainscg

0.998334

0.0015672

0.098452

13

6

6

16

Trainscg

0.998351

0.0016943

0.019254

32

7

8

17

Trainscg

0.999167

0.0019932

0.020350

13

8

12

8

Trainscg

0.994306

0.0016154

0.155355

29

9

8

4

Trainscg

0.992829

0.0008781

0.050850

32

11

11

6

Trainscg

0.997632

0.0008574

0.080555

300

12

15

8

Trainscg

0.998844

0.0005376

0.042757

87

13

17

8

Trainbr

0.999624

0.0013030

0.517751

50

14

14

9

Trainbr

0.997924

0.0005934

0.089754

44

15

11

10

Trainbr

0.993778

0.0006381

0.061952

56

16

13

12

Trainbr

0.992877

0.0007459

0.058452

67

17

5

5

Trainbr

0.991301

0.0013400

0.022575

29

18

15

20

Trainbr

0.997399

0.0008975

0.165450

122

19

14

25

Trainbr

0.998628

0.0019403

0.116350

137

20

12

6

Trainscg

0.919018

0.0069056

0.080658

132

21

9

10

Trainscg

0.927577

0.0079445

0.030357

50

22

9

11

Trainscg

0.902770

0.0048026

0.025055

26

23

23

14

Trainscg

0.895403

0.0786320

0.029658

74

24

12

14

Trainscg

0.942997

0.0229730

0.038558

111

25

9

11

Trainscg

0.963978

0.0098485

0.027520

75

26

16

14

Trainscg

0.964364

0.0036883

0.030589

13

27

19

18

Trainscg

0.975536

0.0088318

0.039597

114