i want to implement recurrent neural network in order to solve optimization problem, can any one help me please with some codes in matlab or any effecient documents.thanks
see:
1- Larry Medsker, Lakhmi C. Jain, Recurrent Neural Networks: Design and Applications, CRC Press, 1999.
2- Gérard Dreyfus, Neural Networks - Methodology and Applications, Springer, 2005.
3- Yoel Tenne, Chi-Keong Goh, Computational Intelligence in Optimization: Applications and Implementations, Springer-Verlag, 2010.
http://www.mathworks.com/matlabcentral/fileexchange/22010-a-very-simple-and-intuitive-neural-network-implementation
If you have Wolfram Mathematica there is a package for Neural Network.
http://www.wolfram.com/products/applications/neuralnetworks/
Bye!
thank you suman, Mariano and ali, but i want knowing if neurodynamic optimization means i will solve mathematical problems using RNN or i will optimize (enhance)the work(functionnality) of RNN
Dear Slama,
With the package I suggest you, you don't have to implement te ANN but only understand how to run the ANN itself. You have to import your dataset, and then run the neural network inside your work.
Mariano.
Dear Najla,
I think you have to make clear your question by giving some considered mathematical problems. For optimizing mathematical problems, metaheuristic algorithms usually are used. For your information, ANNs are usually applied for two purposes: prediction and classification. For both of these purposes, it needs the label class or the value of true prediction, in order to adapt your ANN to your problem. If you want to optimize your mathematical problems, why you do not use a powerful meta heuristic algorithm called Stochastic Fractal Search?
Matlab code:
http://www.mathworks.com/matlabcentral/fileexchange/47565-stochastic-fractal-search--sfs-
Paper:
www.sciencedirect.com/science/article/pii/S0950705114002822
best regards,
Hamid
You can use the Neural Network toolbox GUI in MATLAB using nntool command in command prompt.The trained network can be used as an objective function for solving optimization problems .For example,you can use Neural network as a fitness function in Genetic algorithm .Herein,you will have to make a function which takes in input values and returns the neural network predicted values as output.This function handle(@ )will have to be used in genetic algorithm GUI in MATLAB.
Is this kind of optimization work better than traditional methods like Quasi newton,Alternative minimization projection?
Hi, the following papers can help you:
https://www.researchgate.net/publication/312141846_A_projection_type_steepest_descent_neural_network_for_solving_a_class_of_nonsmooth_optimization_problems
https://www.researchgate.net/publication/324531026_An_efficient_one-layer_recurrent_neural_network_for_solving_a_class_of_nonsmooth_pseudoconvex_optimization_problems
https://www.researchgate.net/publication/283448488_A_non-penalty_recurrent_neural_network_for_solving_a_class_of_constrained_optimization_problems
https://www.researchgate.net/publication/236899596_A_New_Steepest_Descent_Differential_Inclusion-Based_Method_for_Solving_General_Nonsmooth_Convex_Optimization_Problems
https://www.researchgate.net/publication/236899619_A_differential_inclusion-based_approach_for_solving_nonsmooth_convex_optimization_problems
https://www.researchgate.net/publication/236180856_A_recurrent_neural_network_for_solving_a_class_of_generalized_convex_optimization_problems
Hi, I need to know is anyone is working on Hopfield Neural Network for optimization, I need a help how use activation function at the final step.
- Badges
- Science topic