sigmoid(sop) function not numerically safe

[jtlareau]

The current implementation of the sigmoid() function can get a RuntimeWarning overflow in cnn.py at line 29 numpy.exp():
"return 1.0 / (1 + numpy.exp(-1 * sop))"
To avoid/correct the overflow, use:
"return numpy.exp(-numpy.logaddexp(0, -sop))"

[ahmedfgad]

It is just a warning and NumPy is able to handle the situation. It happens when a large number like 1234.1234 is fed to the numpy.exp() function. This does not affect the result.