In this research, we study how artificial neural network can be used in solving both equality and inequality constrained non-linear optimization problems. The procedure involves generating a system of differential equations which must correspond to the given number of intrinsic variables. We proceed to apply this technique to solve some constrained non-linear optimization problems. Using MathCAD14, we were able to generate the corresponding system of differential equations developed using the neural network gradient scheme. We provide the resulting graphical profiles of the solutions and compared it with the analytic solution. Our steps and results indicate that, for the problems tested, the neural network approach can be more straightforward to implement and avoids some of the iterative assumptions of classical methods.
References
[1]
Cichockih, A. (1993) Introduction to Linear and Non-Linear Programming. Addi-son-Valley.
[2]
Chong, E.K.P. and Zak, S.H. (2001) An Introduction to Optimization. 2nd Edition, John Wiley & Sons, Inc. https://onlinelibrary.wiley.com/doi/book/10.1002/9781118033340
[3]
Orponen, P. (2000) An Overview of the Computational Power of Recurrent Neural Networks. In: Hy？tniemi, H., Ed., Proceedings of the 9th Finnish AI Conference (STeP 2000—Millennium of AI): AI of Tomorrow, Symposium on Theory (Vol. 3, pp. 89-96), Espoo, 28-30 August 2000, Finnish Artificial Intelligence Society (Suomen Teko？lyseura), Vaasa.
[4]
Richard, P.L. (1987) An Introduction to Computing with Neural Nets. IEEE ASSP Magazine, 4, 4-22. https://ieeexplore.ieee.org/document/1165576
[5]
Salchenberger, L.M., Cinar, E.M. and Lash, N.A. (1992) Neural Networks: A New Tool for Predicting Thrift Failures. DecisionSciences, 23, 899-916. https://doi.org/10.1111/j.1540-5915.1992.tb00425.x
[6]
Eckmiller, R. and von der Malsburg, C. (Eds.) (1989) Neural Computers. NATO ASI Series F: Computer and Systems Sciences, Vol. 41. Springer-Verlag.
[7]
Cybenko, G. (1989) Approximation by Superpositions of a Sigmoidal Function. MathematicsofControl, Signals, andSystems, 2, 303-314. https://doi.org/10.1007/bf02551274
[8]
Hornik, K., Stinchcombe, M. and White, H. (1989) Multilayer Feedforward Networks Are Universal Approximators. NeuralNetworks, 2, 359-366. https://doi.org/10.1016/0893-6080(89)90020-8
[9]
Luenberger, D.G. (1973) Introduction to Linear and Non-Linear Programming. Addison-Wesley.
[10]
Luenberger, D.G. (1984) Linear and Non-Linear Programming. Addison-Wesley.
[11]
Cichocki, A. and Unbehauen, R. (1993) Neural Networks for Optimization and Signal Processing. John Wiley & Sons, Inc.
[12]
Cichocki, A., Unbehauen, R., Weinzierl, K. and H？lzel, R. (1996) A New Neural Network for Solving Linear Programming Problems. EuropeanJournalofOperationalResearch, 93, 244-256. https://doi.org/10.1016/0377-2217(96)00044-6
[13]
Valeri, M.M. and Nicholas, G.M. (2000) Neural Networks for Solving Constrained Optimization Problems. 4th WSEAS Multi-Conference on Circuits, Systems, Communications and Computers (CSCC’2000), Vouliagmeni (Athens), July 10-15, 1351-1359.
