Table 2. Statistical measures and performances of four learning algorithms on the enzymatic synthesis of betulinic acid ester.



Learning algorithm

The best architecture

Training data

  

Testing data

RMSE

R2

AAD

 

RMSE

R2

AAD

Quick Propagation (QP)

4-9-1

0.0335

0.9999

0.0647

 

0.6279

0.9961

1.4478

Incremental Backpropagation (IBP)

4-9-1

1.1738

0.9900

2.0796

 

1.1469

0.9871

2.2973

Batch Backpropagation (BBP)

4-7-1

0.9582

0.9933

1.7536

 

0.8096

0.9935

1.8932

Levenberg-Marquardt (LM) 

4-8-1

0.2909

0.9993

0.5402

 

1.4214

0.9801

3.2160

 


Supported by UNESCO / MIRCEN network