Scribd
Upload
45 views

RBF Elm PNN-2020

The document discusses three types of neural networks: Radial Basis Function networks (RBF), Extreme Learning Machines (ELM), and Probabilistic Neural Networks (PNN). It provides details on RBF networks, including their structure, use of radial basis functions, and training algorithms like newrbe and newrb in MATLAB. It also briefly introduces ELM networks and indicates it will discuss implementing RBF and ELM networks from scratch.

Uploaded by

youssef hossam
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
45 views

RBF Elm PNN-2020

The document discusses three types of neural networks: Radial Basis Function networks (RBF), Extreme Learning Machines (ELM), and Probabilistic Neural Networks (PNN). It provides details on RBF networks, including their structure, use of radial basis functions, and training algorithms like newrbe and newrb in MATLAB. It also briefly introduces ELM networks and indicates it will discuss implementing RBF and ELM networks from scratch.

Uploaded by

youssef hossam
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 24

RBF, ELM and PNN

Neural Networks
2020
Dr. Mohamed Waleed Fakhr
m
f (x) =  wii (x)
i =1

Radial basis function RBF Model

y
Linearly
Output
weighted
Units
output

w1 w2 wm
• Decomposition
Hidden
Units
1 2 m • Feature Extraction
• Transformation

Input Feature Vector


x = x1 x2 xn
m
f (x) =  wii (x)
i =1

RBF Model
y
Linearly
Output
weighted
Units
output

w1 w2 wm
• Decomposition
Hidden
Units
1 2 m • Feature Extraction
• Transformation

Input Feature Vector


x = x1 x2 xn
m
f (x) =  wii (x)
i =1
Radial Basis Function Networks as
Universal Aproximators

w1 w2 wm With sufficient number of


Hidden radial-basis-function units,
Units
1 2 m it can also be a universal
approximator.

x = x1 x2 xn
m
f (x) =  wii (x)
i =1
Non-Linear Models: we adapt the weights and the
RBF parameters

Weights
m
f (x) =  wii (x)
i =1
m
f (x) =  wii (x)
i =1

Radial Basis Functions

Three parameters
for a radial function:
i(x)= (||x − xi||)
⚫ Center xi
⚫ Distance Measure r = ||x − xi||
⚫ Shape 
− r2

 (r ) = e 2 2
  0 and r  

Gaussian Basis Function (=0.5,1.0,1.5)

 = 1.5
 = 1.0
 = 0.5
In Matlab

⚫ newrbe: uses all the training patterns as


Gaussian centers. Calculates the weights by
solving a least squares simple problem
⚫ newrb: The design method of newrb is
similar to that of newrbe. The difference is
that newrb creates neurons one at a time.
⚫ At each iteration the input vector that results
in lowering the network error the most is
used to create a neuron.
newrb vs newrbe

⚫ The error of the new network is checked, and


if low enough newrb is finished. Otherwise
the next neuron is added. This procedure is
repeated until the error goal is met or the
maximum number of neurons is reached
⚫ We will do our own implementation of the
RBF neural network and the Extreme
Learning Machine neural network (ELM)
which appeared in 2012.

You might also like