First you would need to generate some sample data, feeding some randomly distributed values into your "blackbox" and getting the output.
Then use the obtained set of input/outputs to train the neural-network and once the error goes below your acceptable limit, you could use it to replace the original blackbox.
In reply to Re: Reverse engineering a formula...
by salva
in thread Reverse engineering a formula...
by devnul
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |