Вы находитесь на странице: 1из 9

Computational Intelligence

A Biological Approach to Artificial Neural Network

network is a crude electronic model based on


neural structure of the brain. The brain stores
information as patterns utilizes these patterns and
solves problems. This paper species that how the
human brain learns? And also that defines the
essential

ABSTRACT:

Computational Intelligence is a branch of


study of artificial Intelligence. CI is a set of
Features of neurons and their interconnections.
machine capabilities which allowed a machine to
We then typically program a computer to simulate
use acquired knowledge and to solve new task
these features. In this paper we have shown
efficiently in the presence of new condition of their
theoretically neural network versus conventional
performance. Neural network is a technique of
computers. Neural networks are based as they are
artificial intelligence i.e., computational
on a crude low level model of biological neural
intelligence. It is on information processing
systems. In future, the development of this
paradigm that is inspired by the way biological
neurobiological modeling may lead to intelligent
nervous system, such as the brain, process
genuinely computers.
information. A neural network is an
interconnected group of biological neurons in
modern uses that constituted of artificial
This paper is discussed the following points:-
neurons. In general a neural network is composed
of groups or groups of physically connected or 1. Artificial neural network
functionally associated neurons. A single neuron
2. Neural Networks versus Conventional
can be connected to many other neurons and the
Computers
total number of neurons and connection in a
network can be extremely large. Artificial neural 3. Applications of ANN
4. Limitations of Neuro- Fuzzy Systems techniques and traditional logic based artificial
intelligence techniques. Theoretical and
5. Conclusion
computational neuroscience is the field concerned
with the theoretical analysis and computational
modeling of biological neural system. Neural
networks are based as they are on a crude low
level model of biological neural systems. In future,
the development of this neurobiological modeling
may lead to intelligent genuinely computers.

2. Artificial neural network:

1. Computational Intelligence: The term artificial neural


network encompasses a wide array of
It is a fairly new name covering a computational tools loosely patterned after
fairly new field. Computational Intelligence is a biological processes. Physically, all artificial neural
branch of study of artificial Intelligence. CI is a set networks interconnected assemble of
of machine capabilities which allowed a machine mathematically simple computational elements.
to use acquired knowledge and to solve new task These computational elements a very limited
efficiently in the presence of new condition of their amount of local memory and perform rudimentary
performance. Computational intelligence is the mathematical operations on a data passing
study of the design of intelligence agents. An through them. A neural network is a network
‘agent’ is something that acts in an environment capable of information processing, in which a large
such as mobile, robot, a web Crawler, an number of relatively simple information
automated medical diagnosis system.Some processing units are connected together and in
techniques within computational intelligence are which these units communicate with each other by
often counted as artificial intelligence relatively simple signals. An artificial neural
techniques(i.e., genetic algorithm, neural network ) network also called a simulated neural network or
there is a cleared difference between these commonly just neural network (NN). It is on
information processing paradigm that is inspired
by the way biological nervous system, such as the
(a) A biological neural network is a plexus of
brain, process information.
connected for functions related neurons in the
It works along two directions. Firstly, it peripherial nerves system or the central nervous
deals with realizing the anatomy and functioning system. In the field of neuro science, most often
of the human brain in order to define models refers to a group of neurons from a nervous
adherent to its physical behaviour. Secondly, it system.
aims at computing on the basis of the parallel and
distributed structure of the brain.
(b) Artificial neural networks were designed to
model the same properties of biological neural
networks.

In general a neural network is composed of groups


or groups of physically connected or functionally
associated neurons. A single neuron can be
connected to many other neurons and the total
number of neurons and connection in a network
can be extremely large.

2.1 Analogy to Brain:

Artificial neural network is a crude electronic


Fig: A view of an artificial neural network model based on neural structure of the brain. The
brain stores information as patterns utilizes these
A neural network is an interconnected group of
patterns and solves problems. Behind this power
biological neurons in modern uses that constituted
of brain lies 100 billion neurons co0nnected with
of artificial neurons. The term neural network
specifies the two distinct concepts:
200,000 other neurons.

A Simple Neuron Basic Artificial Neuron.

A biological neuron receives inputs from other


sources and synapses of other neurons, combines
them in some way, performs a generally non-linear
operation on the result and then outputs the final
result. Likewise the artificial neurons simulate the
four basic functions of natural neurons. Various
inputs to the network are represented by the
mathematical symbol, x (n). Each of these inputs
are multiplied by a connection weight w(n).

2.2 How the Human Brain Learns?


Much is still unknown about how the brain
learns itself to process information, so theories
around. In the human brain, a typically neuron
collect signal from other through a long, thin stand
known as an axon, which splits into thousands of
branches, a structure called a synapse converts the
activity from the axon into electrical effects that
inhibit or excit activity from the axon into
electrical effects that inhibits or excite activity in
the connected neurons. When a neuron receives Neural model
excitatory input that is sufficiently large compared
An Artificial neuron is device with many inputs
with its inhibitory input; it sends a spike of
and one output. The neuron has modes of
electrical activity down its axon. Learning occurs
operation, the training mode and using mode. In
by changing the effectiveness of the synapses so
the training mode, the neurons can be treated for
that the influence of one neuron on another
particular inputs patterns. In the using moal when
changes.
are taught input pattern is detected at the input;
its associated output becomes the current output.

2.3 Form Human Neurons to Artificial


Neuron:

We conduct these neural networks by first Neuron


trying to deduce the essential features of neurons
and their interconnections. We then typically
program a computer to simulate these features.
However because our knowledge of neurons is
Inputs x1, x2, x3 Teaching Unit
incomplete and our computing power is limited,
our models are necessary goes idealization of real
networks ofneuronsq.
3. Neural Networks versus Conventional Neural networks take a different approach to

Computers: problem solving than of conventional


computers.conventional computers use an
Computers are not models of the brain (even alogrith approaches that is the computer follows
though it is possible to describe a logical process as the set of instruction in order to solve the
a computer program or to simulate a brain using a problems.unless the specific steps that the needs to
computer) as they were not created with that follow are known the computer .the computer
purpose in mind. cannot solve the problem that restrict the problem
solving capability conventional computer to
problem that we already understand and know
how to solve.but computer would more useful if
they could do things that do not exactly now how
to do .

Neuron networking process in similar way, the


human brain does. The networking is composing

Fig: A neural network of large number of highly interconnected neurons

is an interconnected group of nodes, akin to the working in parallel to solve specific problem.

vast network of neurons in the human brain. Neural networks learns by examples.they cannot
be programmed to perform specific task .the
example must be selected carefully otherwise
useful time is wasted or even worse the networks
might be functioning incorrectly.the disadvantage
is that because the network finds out here to solve
the problem by itself, its operation can be
unpredictable. On the other hand convectional
computer user cognitine approach to problem
solving, the way that problem is to solve must be
known and stated in small unambiguous
instruction. Neural network and conventional
algorithmic computers are not in competition but
complement each other. There are tasks are more of the human body and recognizing disease from
united to algorithmic approach like arithmetic various scans(eg. Cardiograms, CAT scan,
opertion and takes that are more suted to nudel ultrasonics scans etc.). Neural networks are ideal
networks. Even more , a large number of tasks in recognizing disease suing scans since there is no
require system that use a combination of the need to provide a specific algorithim on how to
approaches. (Normally a conventional computer is identify the disease neural network learn examples
used to supervise neural network in a order to so the details of how to recognize the disease are
perform maximum efficiency). not needed what is needed is asset of examples is
not as important as the ”quality”. The examples
need to be selected very carefully is the system is

4. Applications of ANN, Detection of perform reliably and efficiently.

medical phenomenon:

A Variety of health related intriceses (a Instant physician:


combination of heartrate, levels of various
An application development in the mid-
substances in the blood respiration rate) can be
1980s called the”instant physician” trained an
monitor.The onset of a particular medical
auto associative memory neural network to store a
condition could be associated with the very
leere number of medical records. Each of which
complex(eg. Nonlinear and interactive
includes information on symptoms diagnosis and
combinations of changes on a subset of the
treatment for a particular case. After training the
variables being monitored.) Neural networks have
net can be presented with input consisting of a set
been used to recognition this presctive pattern, so
of symptoms: it twice then find the fuel tired
that appropriate treatment can be prescribed.
pattern that represents the ‘best’ diagnosis and
treatment.

Neural networks in medicine: Adsorption water treatment:

Artificial neural network is currently a Artificial neural network can be applied to


“hot” research area in medicine and it is believed the literature data pertaining to adsorption batch
that they will receive extensive application to studies to develop and validate a model that can
biomedical system in the next few years. At the predict the pollutant Removal Efficiency (PRE).
moment, the research is mostly on modeling parts
Adsorption is one of the established unit estimate the quantity before using it for the
operations used for the treatment of contaminated purpose of updating. Accurately to estimate the
water i.e., raw water and or waste water. The quantity before using it for the purpose of
polluted removal efficiency (PRE) is the most updating.for a finite updating budget; one can
significant output of adsorption studies with either spend most of the budgeton estimating each
respect to waste water treatment. The variation of interative steps very accurately and settle for
PRE depends on several factors such as adsorbent fewer steps; or estimate each step very poorly but
characteristics, contact time, adsorbate use the budget to calculate may iterative steps. In
concentration etc. many problems / techniques, the computations
involved group exponentially with the size of the
5. Limitations of Neuro- Fuzzy Systems:
problem which senders such technique
Neural techniques are executed sequentially and impractical.
difficult to parallelize.
6. Conclusion:
When the quantity of data increases, methods may
suffer a combinatorialExplosion.The learning This paper has presented an overview of
process seems difficult to simulate in a symbolic computational intelligence in the form of artificial
system. neural network with biological applications.
Artificial neural network is a crude electronic
In Perceptron network: model based on neural structure of the brain. The

(i)The output values of a perceptron can brain stores information as patterns utilizes these

take only one of two values (eg. 0 and l) due to patterns and solves problems. In this paper, we

hard- limiter transfer function. have tried to show the various aspects of artificial
neural networks in the form of biological
(ii) Perceptrons can only classify linearly
approach like analogy of human brain and heart
separable sets of vectors else learning will never
rate and also for learning human neuron to
reach a point where all vectors are classified
artificial neurons.
properly.

(iii) In computational approach of neural


networks- When we try to solve a stochastic
optimization problem, are need to approximate.
And hence we must decide on have accurately to
7. References: Cybernetics, Volume SMC-13, Number 5,
September, 1983.
(i) Anderson, James A., "A Simple Neural
Network Generating an Interactive Memory", (iii)http://www.erns1.pn1.gov:2080/docs/cie/neural

Mathematical Biosciences, Volume 14, 1972. /neural.homepahe.html

(ii) Anderson, James A., "Cognitive and (iv)

Psychological Computation with Neural Models", http://www.erns1.pn1.gov:2080/docs/cie/techbrief/

IEEE Transactions on Systems, Man, and NN.techbrief.ht

Вам также может понравиться