I made an equation using a Artificial Neural Network (ANN) using MATLAB NN Toolbox for two input and one output.
While the test output (0.863) close to the experimental value (0.8646), the output from the equation shows completely different.
For your information, i was using the following equation format .
Output, Y = LW (x) +b2
The hyperbolic tangent function x = exp(a)-exp(-a)/(exp(a)+exp(-a) ) where a = ∑{(IW∗Input)+b1}.
any advice and suggestions will be greatly appreciated
Thanks in advance.
There may be some data missing during the simulation period. Some error or garbage collection may be occurred during the simulation. but the simulation results must be near about with the equated result.
May be i need to normalize the value of input and output before training as some other thread with related question suggested.
The normalization of inputs and outputs at the phase of the training process are conducted automatically by the software (MATLAB). In my opinion you must conduct the normalization of inputs and the de-normalization (i.e. the reverse transformation) of the outputs manually if you want to simulate the network in a correct way.
- Badges
- Science topic
- Similar topics
- Computer Communications (Networks)
- Simulators