Neural network output should be in the form of equation? Or number? In Response Surface Methodology (RSM) either a single order or second order equation can be used as fitness function for PSO. But ANN output for PSO?
Dear friend
Which output of ANN you are talking about?
can you explain your question little deeper.
What exactly you are trying to do.
Thank you.
Best regards
Dr. Indrajit Mandal
Dear Sir,
I want the output of ANN network to be given as input to PSO algorithm so tat i can optimize the results
PSO the same as the other optimization approaches, is required an objective function to be minimized, therefor you should create a function based on variables you would like to be minimize. see the PSO help for more information.
predefine a fitness or objective function for PSO.
the output of NN is a prediction of a actual output. so you can define some prediction error between the output and actual output.
this could be the mean square error or other kinds of errors.
if PSO has been implemented to find solution with minimum objective function, you should use a measure for error, MSE as Zhongyi suggested. Ali mentioned this point meticulously. If your version of PSO implemented in a way that search to find solutions with maximum objective function, you should use -1 * MSE (or -1 * any other measure for error) as the PSO objective function.
any can tel me about general relation or step to find fitness function
https://www.researchgate.net/publication/297245624_Particle_Swarm_Optimization_Algorithm_and_its_Codes_in_MATLAB
Research Particle Swarm Optimization: Algorithm and its Codes in MATLAB
https://www.researchgate.net/publication/296636431_Codes_in_MATLAB_for_Particle_Swarm_Optimization
Code Codes in MATLAB for Particle Swarm Optimization
https://www.researchgate.net/publication/280244004_A_comparative_study_of_metaheuristic_optimization_approaches_for_directional_overcurrent_relays_coordination
Article A comparative study of metaheuristic optimization approaches...
I defined MSE function as fitness function . when I call this function it gives error as np.array is not callable. please help me. if any one have code in in python for NN prediction error minimization.
Thanks
- Badges
- Science topic