To export NN into equation from NS, you should know:
- The configuration of NN (architecture), to indicate how inputs interconnected to output
- You know the NN parameters such as weight and bias after the training
- Since NSpred and NS, we do not know which NN configuration used. Except you can break the network file (.net in NSPred or .cht on NS) and convert them into input-output data, NN config, weight and bias parameters.
- Seems that on NS, it use extendable hidden neurons (up to 80) but the NN configuration is unknown
If you can give above information, I am very sure, it is possible to convert them into equation.