Transcript What is a Neural Network An artificial neural

Intelligent
Vision
Systems
ENT 496
Lecture 3
Hema C.R.
Introduction to Neural Networks
Artificial Intelligence
• Humans often mimic biological systems to develop new
technology.
• This emerging field of technology is monikered
“Biomimentic”
• Artificial Intelligence is one such technology
• What is Artificial Intelligence
– It is the science and engineering of making intelligent
machines, especially intelligent computer programs.
– It is related to the similar task of using computers to
understand human intelligence
– AI does not have to confine itself to
methods that are biologically observable.
Hema ENT 496
Lecture 3
2
Artificial Intelligence
Techniques
• Neural Networks
– Depicts human brain
• Fuzzy Logic
– Depicts human behavior
• Genetic Algorithms
– Depicts evolution
• Swarm Intelligence
– Depicts group behavior
Hema ENT 496
Lecture 3
3
What is a Neural Network
• An artificial neural network is an
information processing system that has
been developed as a generalization of the
mathematical model of human cognition
• Neural Network functions in a way
similar to the human brain
• The function of a neural network is to
produce an output pattern when
presented with an input pattern.
Hema ENT 496
Lecture 3
4
Human Brain
• The Brain:
– The Brain is composed of 1011 neurons or nerve
cells of different types.
• The Human Body:
– Has 10 to 100 billion special cells, called neurons.
– Each neuron is approximately connected to 10
thousand other neurons via fine fibers to produce a
complex network.
– This network is called the central nervous system,
for which the control point is the brain.
• Neuron:
– They are specialized tiny cells.
– Less than 100 micron width
one micron is one millionth of a meter.
Hema ENT 496
Lecture 3
5
Biological Neuron
• A biological neuron has three
types of components
– Dendrites
• Receives signal from other neurons
– Soma
(soma)
• Sums the incoming signals
• When sufficient input is received the
cell fires
– Axon
• Impulse signal triggered
is transmitted over the
axon to other cells
Hema ENT 496
Lecture 3
• Connecting point between
a neuron’s axon and another neuron’s
dendrite is called a synapse
6
Artificial Neuron
• The artificial neuron mimes the
characteristics of the biological neuron.
• The artificial neuron has a set of ‘n’ inputs
xi, each representing the output of another
neuron.
• The inputs are collectively referred as X.
Each input is weighed before reaching the
main body of the processing element by
the connection strength or the weight
factor or simply weight, analogous
to the synaptic strength.
Hema ENT 496
Lecture 3
7
• Each signal is multiplied by an
associated weight w1, w2, w3,…,wn before
it is applied to the summing block.
• In addition, the artificial neuron has
– a bias term w0,
– a threshold value ‘’ that has to be reached
or extended for the neuron to produce a
signal
– a nonlinear function ‘F’ that acts on the
produced signal ‘net’ and
– an output ‘Y’
• It should be noted that the input to the bias neuron is
assumed to be 1.
Hema ENT 496
Lecture 3
8
Basic Neuron Model
1
X0
w0
x1
X1
w1
x2
X2
w3
x3
.
.
.
xn
w2
X3
∑
NET
F (NET)
Y = OUT
Summing
Block
wn
Xn
Hema ENT 496
Lecture 3
9
Transfer function of the basic
neuron model.
y =F(NET)
where,
Hema ENT 496
Lecture 3
NET = w0 + x1 w1 + x2 w2 + x3 w3 + … + xn wn
10
ACTIVATION FUNCTIONS
The NET signal is usually further
processed by an activation function F
to produce the neuron’s signal, OUT.
This may be a simple linear function,
OUT = F(NET)
Where f is a constant, a threshold function,
OUT = 1 if NET > T
OUT = 0 otherwise
Hema ENT 496
Lecture 3
11
ARCHITECTURE OF A
NEURAL NETWORK
The arrangement of neurons into layers
and the connection pattern within and
between the layers is known as
network architecture.
Hema ENT 496
Lecture 3
12
Input Layer
The neurons in this layer receive the
external input signals and perform no
computation, but simply transfer the
input signals to the neurons in another
layer.
Hema ENT 496
Lecture 3
13
Hidden Layer
The layer of neurons that are
connected in-between the input layer
and the output layer is known as
Hidden layer.
Hema ENT 496
Lecture 3
14
Output Layer
The neurons in this layer receive
signals either from the neurons in the
input layer or the hidden layer.
Hema ENT 496
Lecture 3
15
Neural Nets Classification
• Single layer
• Multilayer networks.
In determining the number of layers, the
input layer is not counted as a layer, because
it does not perform any computation.
The number of layers in a net can be defined
as the number of layers of weighted
interconnection links between the various
layers.
The architectures of a single layer is
shown in Figure.
Hema ENT 496
Lecture 3
16
Single Layer Net
• A single layer net consists of one layer
of connection weights.
• The net consists of a layer of units
called input layer, which receive signals
from the outside world and
• A layer of units called output layer
from which the response of the net can
be obtained. This type of network can be
used for pattern classification problems.
Hema ENT 496
Lecture 3
17
Architecture of a single
layer net
x1
y1
X1
w11
Y1
wi1
wn1
w1j
xi
Xi
wij
yj
Yj
wnj
w1m
wim
xn
Xn
Hema ENT 496
Lecture 3
wnm
ym
Ym
18
TRAINING OF ARTIFICIAL
NEURAL NETWORKS
The basic questions regarding the
learning process are:
1) How do we learn?
2) Which is the most efficient process for
learning?
3) How much and how fast can we learn?
4) What are the hurdles in the learning
process?
Hema ENT 496
Lecture 3
19
Learning is nothing but
training
It is a process in which the network
adjusts its parameters, the synaptic
weights in response to input stimuli so
that the actual output response
converges to the desired output
response. When the actual output
response is the same as the desired
one, the network has completed the
learning phase and the network
has acquired knowledge.
Hema ENT 496
Lecture 3
20
Network Training
Learning or training algorithms can
be categorized as:
1. Supervised training
2. Unsupervised training
3. Reinforced training.
Hema ENT 496
Lecture 3
21
APPLICATIONS OF NEURAL
NETWORKS
There have been many impressive
demonstrations of artificial neural
networks. A neural network can discover
the distinguishing features needed to
perform
a
classification
task.
Classification is the assignment of each
object to a specific class, which is an
important aspect in image
classification.
Hema ENT 496
Lecture 3
22
Neural networks have been used
successfully used in a large number of
classification tasks which includes:
1. Recognition of printed or handwritten
characters.
2. Classification of SONAR and RADAR
signals.
3. Signal Processing
4. Speech Recognition
5. Medicine
6. Intelligent Control
7. Function Approximation
8. Pattern analysis
Hema ENT 496
Lecture 3
23
Introduction to
Neural Networks
End of Lecture 3