Download Implementation of parallel Optimized ABC Algorithm

Implementation of Parallel Optimized ABC Algorithm with SMA
Technique for Garlic Expert Advisory System
First Author
N.Thirupathi Rao
M.Tech (CST with AI & R)
Andhra University
Visakhapatnam
Email: [email protected]
Second Author
Prof. M Surendra Prasad Babu
Dept. of CS & SE
Andhra University
Visakhapatnam
Email: [email protected]
‘Garlic Expert Advisory System’.
Abstract:
The present paper deals with
This system is aimed at identifying
the development of expert systems
the
using machine learning techniques
management
to advice the farmers in villages
production to advise the farmers in
through online. An expert system
the villages on line to obtain
is a computer program, with a set
standardized yields. This advisory
of rules encapsulating knowledge
system is designed by using JSP as
about a particular problem domain.
front end and MYSQL as backend.
Machine Learning is a mechanism,
Key words: Expert Systems,
diseases
and
in
disease
garlic
crop
used in the development of Expert
Machine
Learning,
systems, concerned with writing a
Algorithm,
SMA
computer
that
Garlic Crop, Optimization, JSP &
with
MYSQL.
program
automatically
improves
experience. ABC Algorithm was
Introduction:
considered as base and proposed a
Expert Systems:
ABC
Technique,
new technique known as ‘Shared
Expert systems can be defined as a tool
Memory
for
Technique)
Algorithm
Architecture’
to
design
named
(SMA
a
information
knowledge.
new
Expert
implementations
Parallel
generation
System
from
(ES)
automatically
Optimized ABC Algorithm with
perform tasks for which specially
SMA
this
trained or talented people required.
ABC
Expert systems are most common in a
Algorithm we developed a new
specific problem domain, traditional
Parallel
Technique.
Using
Optimized
application and subfield of artificial
1
intelligence.
A
wide
variety
of
Maharashtra
and
Bihar
are
the
methods can be used to study the
premium producers of Garlic in India.
performance of an expert system.
There are about six popular varieties in
Expert systems might have learning
garlic. Garlic has germanium in it.
components but a common element is
Germanium is an anti-cancer agent,
that once the system is developed, it is
and garlic has more of it than any other
proven by placing it in the same real
herb. Another benefit of garlic is it
world problem solving situation as the
helps
human SME, typically which is an aid
pressure.
to human workers or a supplement to
problems with low or high blood
some
The
pressure, garlic can help equalize it. In
sequence of steps followed to reach a
addition to all these health benefits,
conclusion is dynamically synthesized
garlic is packed with vitamins and
with each new case. It is not explicitly
nutrients. Some of these include
programmed when the system is built.
protein, potassium, Vitamins A, B, B2
Problem solving is accomplished by
and C, Calcium, Zinc and many others.
applying specific knowledge rather
Machine Learning:
information
system.
regulate
So
the
body's
whether
blood
you
have
Machine Learning is a mechanism
than specific technique. This is a key
idea in expert systems technology.
that concerned with writing a computer
Garlic:
program that automatically improves
Garlic is one of the most commonly
with experience. It is a very young
used vegetables in India. Garlic is also
scientific discipline whose birth can be
known as Lassan and its botanical
placed in the mid-seventies. The First
name is Allium sativa Linn. It belongs
Machine Learning Workshop was
to the Lilliaceae family and is known
taken place in 1980 at Carnie-Mellon
by several many names in different
University
parts of India. Its Sanskrit name is
machine learning is to program the
Lashuna. Garlic is called as Velluri in
computers such that to use example
Telugu.
is
data or past experience to solve a given
cultivated in all over India but
problem. There were many successful
Rajasthan,
applications of machine learning exists
Even
though
Karnataka,
garlic
TamilNadu,
2
(USA).
The
goal
of
today, including systems that analyze
optimization problems. It is very
past sales data to predict customer
simple and flexible when compared to
behavior, recognize faces or spoken
the other Swarm Based algorithms
speech, optimize robot behavior so that
such as Particle Swarm Optimization
a
using
(PSO). It does not require external
extract
parameters like mutation and crossover
task
can
minimum
be
completed
resources,
and
knowledge from bioinformatics data.
rates, which are hard to determine in
ABC Algorithm:
prior. The algorithm combines local
The Artificial Bee Colony (ABC)
search methods with global search
Algorithm
[1, 3, and 4]
is a meta-heuristic
methods and tries to attain a balance
algorithm for numerical optimization.
between exploration and exploitation.
Meta-heuristics
high-level
Researchers have come up with several
strategies for exploring search spaces.
real-world applications for the ABC
Many
algorithm.
are
meta-heuristic
algorithms,
inspired from nature, are efficient in
Proposed System:
solving
optimization
Here a web based application of
problems. ABC algorithm is motivated
Expert advisory system for garlic crop
by the intelligent foraging behavior of
is developed by using ABC Algorithm
[6, 7]
as base and modified this ABC
was first proposed by Karaboga in
Algorithm. The proposed Architecture
2005 for unconstrained optimization
is as follows:
numerical
honey bees. The ABC algorithm
problems. Subsequently, the algorithm
has been developed by Karaboga and
Basturk and extended to constrained
optimization problems. Improvements
to the performance of the algorithm
and a hybrid version of the algorithm
have been also been proposed. The
Proposed Algorithm:
ABC algorithm is a swarm-based
In general, parallel architectures may
algorithm good at solving unimodal
use either a shared memory or a
and
message
multimodal
numerical
3
passing
mechanism
to
Step.3. Steps 4 to 10 are carried out in
parallel at each processor Pr.
Step.4. For each solution in the local
memory Mr of the range processor Pr,
determine a neighbor.
Step.5. Calculate the optimization for
the solutions in Mr.
Step.6. Place the onlookers on the food
sources in Mr and improve the
corresponding solutions (as in step 4).
Step.7. Determine the abandoned
solution (if any) in Mr and replace it
with a new randomly produced
solution.
Step.8. Record the best local solution
obtained till now at Pr.
Step.9. Copy the solutions in Mr to the
corresponding slots in S.
Step.10. Repeat steps 4 to 9 until MCN
cycles are completed.
Step.11. Determine the global best
solution among the best local solutions
recorded at each processor.
communicate in between the multiple
processing
elements.
Parallel
metaheuristic algorithms have been
developed for both these kinds of
architectures.
A
parallel
implementation of the algorithm is
designed for an optimized shared
memory
architecture,
which
overcomes these dependencies. The
entire colony of bees is divided equally
among the available processors. Each
processor has a set of solutions in a
local memory. A copy of each solution
is also maintained in a global shared
memory. During each cycle the set of
bees at a processor improves the
Database Generation:
solutions in the local memory. The
output optimization can be taken as the
In this section, the setup for production
rules in the knowledge base is presented.
Generally the rules are of the form,
total number of symptoms matching
divided by the total number of
Rule 1: S1=1,S2= 0,S3= 0,S4= 0,
S5=0,S6= 1,S7= 0,S8=1, S9= 0,S10=
0,S11= 0,S12= 0
Resultant disease may be D1
symptoms in the system. At the end of
the cycle, the solutions are copied into
the corresponding slots in the shared
memory by overwriting the previous
Rule 2: S1= 1,S2=1 ,S3= 0 ,S4= 0, S5=
0,S6= 0 ,S7=1,S8= 0 ,S9= 0 ,S10= 0
,S11=0,S12= 1Resultant disease may be
D2
copies. The solutions are thus made
available to all the processors.
Step.1. Generate SN initial solutions
randomly and evaluate them. Place
them in the shared memory S.
Step.2. Divide the solutions equally
among p processors by copying SNp
solutions to the local memory of each
processor.
Rule 3: S1= 0,S2= 1 ,S3= 0 ,S4= 0 , S5=
1,S6= 1 ,S7= 0,S8= 0 ,S9= 0 ,S10=1
,S11=0 ,S12= 0 Resultant disease may
be D3.
4
D
i
s
e
a
s
e
S S S S S S S S S S S S S C
1 2 3 4 5 6 7 8 9 1 1 1 1 u
0 1 2 3 r
e
D
1
D
2
D
3
D
4
D
5
D
6
1 1 0 0 0 0 0 0 0 0 0 0 0 C
1
0 0 1 0 1 1 0 0 0 0 0 0 0 C
2
0 0 0 1 0 0 1 1 0 0 0 0 0 C
3
0 0 1 0 0 0 1 1 0 0 1 0 0 C
4
0 0 0 0 0 0 0 0 1 1 0 1 0 C
5
0 0 0 0 0 0 0 0 0 0 1 1 1 C
6
Fig: 3. Displaying advice to the end user
Conclusions:
In the proposed system, A Parallel
Implementation
of
Artificial
Colony
Bee
Optimized
(ABC)
Algorithm was developed which gives
better
results
implementation
Table:1 Database format
compared
of
general
to
ABC
Algorithm. In the present investigation
Test Results:
it
was
found
that,
the
Parallel
Optimized ABC Algorithm with SMA
Technique gives a better optimization
compared
with
general
ABC
Algorithm. The algorithm used in the
present system can be treated as quite
effective; in most of the cases it finds a
Fig:1. Selection of Symptoms
solution which represents a good
approximation to the optimal one.
Its main emphasis is to have a well
designed interface for giving garlic
plant related advices and suggestions
in the area to farmers by providing
facilities
Fig:2. Selection of Symptoms
like
online
interaction
between expert system and the user
without the need of expert all times.
5
By the thorough interaction with the
Optimization
users
Advances
and
beneficiaries
the
Problems,
in
Soft
LNCS:
Computing:
functionality of the System can be
Foundations of Fuzzy Logic and Soft
extended further to many more areas in
Computing, Vol: 4529/2007, pp: 789-
and around the world.
798, Springer- Verlag, 2007, IFSA
References:
2007. doi: 10.1007/978-3-540-72950-
1. B.Basturk, Dervis Karaboga, An
1_77
Artificial
Bee
Algorithm
for
Optimization,
Colony
(ABC)
5. D. Karaboga, B. Basturk Akay,
Numeric
function
Artificial Bee Colony Algorithm on
IEEE
Swarm
Training Artificial Neural Networks,
Intelligence Symposium 2006, May
Signal
12-14, 2006, Indianapolis, Indiana,
Communications Applications, 2007.
USA.
SIU 2007, IEEE 15th. 11-13 June
2. D. Karaboga, B. Basturk, A
2007,
Powerful and Efficient Algorithm for
10.1109/SIU.2007.4298679
Numerical
Function
6. D. Karaboga, B. Basturk Akay, C.
Artificial
Bee
Algorithm,
Optimization:
Colony
Journal
of
Processing
Page(s):1
-
and
4,
doi:
(ABC)
Ozturk, Artificial Bee Colony (ABC)
Global
Optimization Algorithm for Training
Optimization, Volume: 39, Issue: 3,
Feed-Forward
pp: 459-471, Springer Netherlands,
LNCS:
2007. doi: 10.1007/s10898-007-9149.
Artificial Intelligence, Vol: 4617/2007,
3. D. Karaboga, B. Basturk, On The
pp:318-319, Springer-Verlag, 2007,
Performance Of Artificial Bee Colony
MDAI 2007. doi: 10.1007/978-3-540-
(ABC)
73729-2_30
Algorithm,
Computing,
January
Volume
2008,
Applied
8,
Pages
Issue
Soft
Neural
Modeling
Networks,
Decisions
for
1,
7. J. Vesterstrøm, J. Riget, Particle
687-697.
swarms extensions for improved local,
doi:10.1016/j.asoc.2007.05.007
multimodal and dynamic search in
4. D. Karaboga, B. Basturk, Artificial
numerical optimization, M.Sc. Thesis,
Bee
May 2002.
Colony
(ABC)
Optimization
Algorithm for Solving Constrained
6