Rukba Firdose
5th Sem BCA, NCMS
Radial basis function neural network
(RBFNN) is very prominent in data processing. However, improving this technique
is vital for the NN training process. Neural -symbolic integration combines
both the connectionist system and symbolic artificial intelligence (AI)
systems. Radial basis function (RBF) networks typically have three layers: an
input layer .a hidden layer with a non -linear RBF activation function and a
linear output layer. It is used for the approximation and recognition pattern. And
also multi quality function. The input layer is made up of some node (sensory
units) that connect its network to its environment. The secondary layer
consisting of hidden layer which has hidden units. Apply a non linear transformation
from the input space to its hidden spaces. The output layer is linear it is
basically designed to supply the input from the source node to the activation
function and produce the output and it is work as supervised learning layer.
From the output layer we can calculate the errors as well.
In single perceptron / multi -layer
perceptron (MLP), we only have linear separability because they are composed of
input layers (some hidden layers in MLP).
For example:
The local modelling capability of RBF is demonstrated on the logical operators AND and OR. The examples used are demonstrative for the structure and physical meaning of the weights of the RBF network. Two datasets of four training objects operators AND and OR. The datasets and the network topology are graphically depicted . the AND operator yields a positive output on inputs of identical sign, whereas the OR-operator responds positivity on any positive input.
Different learning algorithm may be used for learning the RBF network parameters. We describe three possible methods for learning , spreads and weights.
The second or hidden layer performs a non-linear mapping from the input space into a (usually) higher dimensional space in which the patterns become linearly separable.
ADVANTAGES OF RBFNN:
Easy design
Good generation
Strong tolerance to input noise
Online learning ability
RBF networks make it very suitable to design flexible control systems.
DISADVANTAGES OF RBFNN
Although the training is faster in RBF network but classification is slow in comparison to multi layer perceptron due to fact that every node in hidden layer have to compute the RBF function for the input sample vector during classification.
A function radial basis if its output depends on (is a non-increasing function of) the distance of the input from a given stored vector.
Click below to learn more
No comments:
Post a Comment