Each algorithm can be understood as a function. A network is also a function thus a network can learn to work in place of an algorithm, however, neuralnet will return roughly results instead of exact results thus it can be called algorithm approximator.
The Benefit
Using neuralnet as algorithm approximator has a big benefit as:
Classic algorithm takes intensive amount of time to give output Neuralnet takes a tiny amount to time for a feed to give output The Drawback
Neuralnet as algorithm approximator has drawback too
Classic algorithm may/or may not give exact results Neuralnet always gives approximate results, never exact.
