Title | ||
---|---|---|
Connectionist models for approximate solutions of non-linear equations in one variable |
Abstract | ||
---|---|---|
In this paper, six neural network models for the computation of an approximate real root of a given non-linear equation are proposed. The models are recurrent with one or more layers. The delay and the feedbacks are automatically taken care by the network itself. The proposed networks are: (1) Bisect net for Bisection method, (2) n-sect net for n-section method, (3) RF-net for regula falsi method, (4) Δ2-net for Attkin's method, (5) NR-net for Newton-Raphson method and (6) K-net for Kizner method. Some of the neurons in the proposed networks use the given function as their activation function. For a general purpose hardware realization, it is possible to replace each such neuron by a composite network such as an MLP (multi-layer perceptron) or a RBF (radial basis function) subnetwork. Our simulation results are obtained using a trained MLP for such neuron. |
Year | Venue | Keywords |
---|---|---|
2003 | Neural Parallel & Scientific Comp. | n-section method,composite network,radial basis function,connectionist model,neural network model,newton-raphson method,proposed network,approximate solution,regula falsi method,activation function,bisection method,non-linear equation,kizner method,neural networks,linear equations,numerical methods |
Field | DocType | Volume |
Bisection method,Mathematical optimization,Nonlinear system,Radial basis function,Activation function,Computer science,Algorithm,False position method,Numerical analysis,Artificial neural network,Perceptron | Journal | 11 |
Issue | Citations | PageRank |
3 | 0 | 0.34 |
References | Authors | |
3 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Srimanta Pal | 1 | 242 | 32.13 |
Nikhil R. Pal | 2 | 4464 | 417.55 |