Neural Network Implementation for Function Approximation
- Login to Download
- 1 Credits
Resource Overview
The primary functions utilized for implementing function approximation via neural networks are: NEWRB - designs a radial basis function neural network, and SIM - performs simulation of the radial basis function network. The NEWRB function automatically determines the optimal number of neurons and centers while adjusting weights through least-squares optimization. The SIM function executes network inference by calculating radial basis function activations and weighted summations.
Detailed Documentation
In this implementation, the network-based function approximation primarily utilizes the following key functions: NEWRB - a method for designing radial basis function neural networks that employs an iterative growth strategy to add neurons until meeting the specified error goal; SIM - a method for simulating radial basis function networks that computes network outputs based on input patterns and learned parameters. Through these functions, we can effectively achieve the objective of function approximation using neural networks. The implementation typically involves first creating the network with NEWRB using training data, then applying the trained network to new inputs with SIM for prediction or approximation tasks.
- Login to Download
- 1 Credits