To "run" a net means to use a saved net. There is no reason why a saved net will yield different results when it is repeatedly run on the same data.
However, if you mean to repeatedly "train" and run, then different results are to be expected because of the following defaults:
1. The data is randomly divided into trn/val/tst data.
2. Initial weights are chosen randomly.
Therefore, if you want to repeat training results, you should reset the random number generator to the same starting state before the first call to init, configure or train. Search on
greg rng(0)
Thank you for formally accepting my answer
Greg
P.S. If you are using obsolete functions like newfit, newpr or newff, you should initialize the RNG previous to the creation of the net.
